wireless networks

Giuseppe Bianchi

bianchi@elet.polimi.it

Giuseppe Bianchi

Course outline

ÎPart 1: cellular planning concepts

ÎPart 2: GSM

ÎPart 3: Wi-Fi

ÎGPRS, UMTS (extra classes - TIM)

ÎExtra time?

Giuseppe Bianchi

1
 Wireless communication
ÎEarly wireless communication:
in the 400-900 TeraHertz Band!
Ö150 BC smoke signals (Greece)
Ö1794, optical telegraph, Claude Chappe

ÎWhat is wireless communication:

ÖAny form of communication that does not require the
transmitter and receiver to be in physical contact
ÖElectromagnetic wave propagated through free-space
ÆRadar, RF, Microwave, IR, Optical

Giuseppe Bianchi

types of communication
ÎSimplex
Öone-way communication
Æradio, TV, etc
ÎHalf-duplex:
Ötwo-way communication but not simultaneous
Æpush-to-talk radios, etc
ÎFull-duplex:
Ötwo-way communication
Æcellular phones
ÖFrequency-division duplex (FDD)
ÖTime-division duplex (TDD): simulated full-duplex

Giuseppe Bianchi

2
 Why wireless communication?
ÎUser Mobility
ÎReduced Cost (cheap infrastructure)
ÖCabling very critical
ÖDeveloping nations utilize cellular telephony rather than
laying twisted-pair wires to each home
ÎFlexibility
ÖCan easily set-up temporary LANs
ÆDisaster situations
ÆOffice moves
ÎOnly use resources when sending or
receiving a signal

Giuseppe Bianchi

Why wireless different than

wired?
ÎNoisy, time-varying channel
ÖBER varies by orders of magnitude
ÖEnvironmental conditions affect transmission
ÎShared medium
ÖOther users create interference
ÖMust develop ways to share the channel
ÎBandwidth is limited
Öspectrum allocated by state rules
ÖISM band for unlicensed use

Giuseppe Bianchi

3
 History of wireless
communication
Î 1896: Marconi
Ö first demonstration of wireless telegraphy
Ö tx of radio waves to a ship at sea 29 km away
Ö long wave transmission, high power req. (200 kW and +)
Î 1901: Marconi
Ö Telegraph across the atlantic ocean
Ö Close to 3000 Km hop!
Î 1907 Commercial transatlantic connections
Ö huge ground stations (30 by100m antennas)
Î 1915: Wireless telephony established
Ö NY – S. Francisco
Ö Virginia and Paris
Î 1920 Marconi:
Ö Discovery of short waves (< 100m)
Ö reflection at the ionosphere
Ö (cheaper) smaller sender and receiver, possible due to the invention of the
vacuum tube (1906, Lee DeForest and Robert von Lieben)

Giuseppe Bianchi

History of wireless
communication
Î 1920's: Radio broadcasting became popular
Î 1928: many TV broadcast trials
Î 1930's: TV broadcasting deployment
Î 1946: First public mobile telephone service in US
Ö St. Louis, Missouri
Ö Single cell system
Î 1960's: Bell Labs developed cellular concept
Ö brought mobile telephony to masses
Î 1960’s: Communications satellites launched
Î Late 1970's: technology advances enable
affordable cellular telephony
Ö entering the modern cellular era
Î 1974-1978: First field Trial for Cellular System
Ö AMPS, Chicago

Giuseppe Bianchi

4
 1st generation mobile systems
ÎFirst generation: 1980’s ÎAnalog transmission
ÎSeveral competing standards in Ö Frequency modulation
different countries ÎVarious bands:
Ö NMT (Nordic Mobile Telephone) Ö NMT:
ÆScandinavian standard; adopted in most Æ450 MHz first
of Europe Æ900 MHz later
ÆFirst european system (Sweden, 1981) Ö TACS
Ö TACS (Total Access Communication Æ900 MHz
Systems), starts in 1985 Ö AMPS
ÆUK standard; A few of Europe, Asia,
Æ800 MHz
Japan
Ö AMPS (Advanced Mobile Phone ÎToday still in use in low-
Service) technology countries
ÆUS standard Ö And not yet completely
Ö C-Netz (Only in Germany) dismissed in high-tech
Ö Radiocom 2000 (Only in France) countries
Giuseppe Bianchi

2nd generation mobile systems

Î4 systems
ÖGlobal System for Mobile (GSM)
ÖDigital AMPS (D-AMPS), US
ÖCode Division Multiple Access
(IS-95) – Qualcomm,US
ÖPersonal Digital Cellular
(PDC),Japan
ÎGSM by far the
dominant one
ÖOriginally pan-european
ÖDeployed worldwide
Æ(slow only in US)
ÎBasic bands:
Ö900 MHz
Ö1800 MHz
Æ(Digital Cellular System:
DCS-1800)
Ö1900 MHz
Æ(Personal Communication
System:PCS-1900,US only)
ÎSpecifications for
ÖGSM-400 (large areas)
ÖGSM-800 (north america)

Giuseppe Bianchi

5
 Timing
Î1982: Start of GSM-specification in
Europe (1982-1990)
Î1983: Start of American AMPS
widespread deployment
Î1984 CT-1 standard (Europe) for
cordless telephones
Î1991 Specification of DECT
ÖDigital European Cordless Telephone (today: Digital
Enhanced Cordless Telecommunications)
- ~100-500m range, 120 duplex channels, 1.2Mbit/s data
transmission, voice encryption, authentication
Î1992: Start of GSM operation Europe-
wide
Î1994: DCS-1800
Giuseppe Bianchi

2 ½ generation mobile systems

GSM incremental extension
ÎHigh speed circuit switched data
(HSCSD)
ÆCircuit switched data communication
ÆUses up to 4 slots (1 slot = 9.6 or 14.4 Kbps)
ÎGeneral Packet Radio Service (GPRS)
ÆPacket data (use spectrum only when needed!)
Æ dial-up comparable speed
ÎEnhanced Data-rates for Global Evolution
(EDGE)
ÆHigher data rate available on radio interface (3x)
» Up to 384 Kbps (8 slots)
» Thanks to new modulation scheme (8PSK)
» May coexist with old GMSK

Giuseppe Bianchi

6
3rd generation mobile systems

ÎUMTS (Universal Mobile

TelecommunicationSystem)
ÆITU standard: IMT-2000 (International Mobile
Telecommunication – 2000)
ÆUMTS forum created in 1996
ÆLater on 3GPP forum (bears most of
standardization activities)
ÖWideband CDMA radio interface
ÆBut several other proposals accepted as
“compatible”
ÖRadio spectrum: 1885-2025 & 2110-2200 MHz

Giuseppe Bianchi

History of Wireless Data

Î Early Wireless LAN proprietary products
Ö WaveLAN (AT&T) - the ancestor of 802.11
Ö HomeRF (Proxim)
Æ45% of the home network in 2000; 30% in 2001, … ε% today
ÆAbandoned by major chip makers (e.g. Intel: dismissed in april 2001)
Î IEEE 802.11 Committee formed in 1990
Ö Charter: specification of MAC and PHY for WLAN
Î First standard: june 1997
Æ1 and 2 Mbps operation
Î Reference standard: september 1999
Ö Multiple Physical Layers
Æ2.4GHz Industrial, Scientific & Medical shared unlicensed band
» Legacy; 802.11b/g
Æ5 GHz ISM (802.11a)
Î 1999: Wireless Ethernet Compatibility Alliance
(WECA) certification
Ö Later on named Wi-Fi
Ö Boosted 802.11 deployment!!

Giuseppe Bianchi

7
 WLAN speeds

Î802.11a: PHY for 5 GHz

Î802.11b: higher rate PHY for 2.4

GHz

Î802.11g: OFDM for 2.4 GHz

Î802.11n: ??? (Higher data rate)

ÆLaunched in september 2003
ÆMinimum goal: 108 Mbps (but higher numbers
considered)

Giuseppe Bianchi

Why so much talking about of

802.11 today?
Î802.11: no more “just” a WLAN
ÎHot-spots
ÖWhere the user goes, the network is available: home,
school, office, hotel, university, airport, convention center…
ÖFreedom to roam with seamless connectivity in every
domain, with single client device
ÎMay compete (complement) with 3G for
Wireless Internet access

Which of these two is the

proper (closer) picture
of Wireless Internet and
Mobile Computing?
Which technology is most suited?

Giuseppe Bianchi

8
 WLAN Market - HotSpots
U.S. Commercial Hotspots
2001-2002: exceeded
U.S. Hotspots growth
expectations by 14%
(2002-2006)
4.500 50000
4.000 40000
3.500
30000
3.000
20000
2.500
10000
2.000
1.500 0
2002 2004 2006
2001 2002
Unique U.S. Hotspots
Forecasted Actual

2003: 125.000 regular hotspot US users

End 2006: 9 Million regular hotspot US users
End 2006: 1 Billion dollars revenue predicted from HotSpot operation
Giuseppe Bianchi

The global picture

GPRS, 3G – UMTS Wide Area
< 400 Kb/s Æ Kms
WAN:
everywhere outside of
the hotspots, where
Local Area wireless Internet
Mobile Broadband Internet 802.11/UMTS connection are
IEEE 802.11 (b) provided
> 10 Mb/s Æ 100 m
switching
LAN:
Personal Area collection of secure
“hot spot”
connections, providing
Bluetooth broadband access to
< 800 Kb/s Æ 10 m BT/802.11
the Internet
switching
PAN:
collection of secure
connections between
devices in a
“very” local area

Giuseppe Bianchi

9
 PART 1
Propagation Characteristics of
Wireless Channels

Lecture 1.1
Basic concepts and
terminology

Giuseppe Bianchi

The Radio Spectrum

s (t ) = A cos(2πft + ϕ )
ÎRadio wave
ÖWavelength λ = c/f
ÖSpeed of light c=3x108 m/s
ÖFrequency: f

[V|U|S|E]HF = [Very|Ultra|Super|Extra] High Frequency

λ

f
f = 900 MHz Æ λ = 33 cm
Giuseppe Bianchi

1
 The radio spectrum
ELF <3 KHz Remote control, Voice, analog phone
VLF 3-30 KHz Submarine, long-range
LF 30-300 KHz Long-range, marine beacon
MF 300 KHz –3 MHz AM radio, marine radio
HF 3-30 MHz Amateur radio, military, long-distance aircraft/ships
VHF 30-300 MHZ TV VHF, FM radio, AM x aircraft commun.
UHF 300 MHz - 3 GHz Cellular, TV UHF, radar
SHF 3-30 GHz Satellite, radar, terrestrial wireless links, WLL
EHF 30-300 GHz Experimental, WLL
IR 300 GHz – 400 THz LAN infrared, consumer electronics
Light 400-900 THz Optical communications

Giuseppe Bianchi

Attenuation phenomena for

millimeter waves (EHF)

Impairments due to
-Oxygen
- water vapour

Giuseppe Bianchi

2
 Spectrum Allocation
Î Cellular systems
Ö 400-2200 MHz range (VHF-UHF)
Ö Simple, small antenna (few cm)
Ö With less than 1W transmit power, can cover several floors within
a building or several miles outside
Î SHF and higher for directed radio links,
satellite communication
Ö Large bandwidth available
Î wireless data systems
Ö 2.4, 5 GHz zones (ISN band)
Ö Main interference from microwave ovens
Ö limitations due to absorption by water and oxygen - weather
dependent fading, signal loss due to by heavy rainfall etc.

Giuseppe Bianchi

Higher-lower Frequencies

ÎHigher frequencies:
Ömore bandwidth
Öless crowded spectrum
Öbut greater attenuation through walls
ÎLower frequencies
Öbandwidth limited
Ölonger antennas required
Ögreater antenna separation required
Öseveral sources of man-made noise

Giuseppe Bianchi

3
 Antennas

ÎTransmission and reception are

achieved by means of an antenna
ÎAn antenna is an electrical
conductor or system of conductors
ÖTransmission - radiates electromagnetic energy into space
ÖReception - collects electromagnetic energy from space
ÎIn two-way communication, the
same antenna can be used for
transmission and reception

Giuseppe Bianchi

Antenna Gain
Î Isotropic antenna (idealized)
Ö Radiates power equally in all directions (3D)
Ö Real antennas always have directive effects (vertically and/or
horizontally)
Î Antenna gain
Ö Power output, in a particular direction, compared to that produced in any
direction by a perfect omni-directional antenna (isotropic antenna)
power density at a distance d in the direction of maximum radiation
Directivity D =
mean power density at a distance d
power density at a distance d in the direction of maximum radiation
Gain G =
PT / 4πd 2

Î Directional antennas “point” energy in a

particular direction
Ö Better received signal strength
Ö Less interference to other receivers
Ö More complex antennas
Giuseppe Bianchi

4
 Radiation Patterns
Î Graphical representation of radiation properties
of an antenna
Î Depicted as two-dimensional cross section

y y z

simple
x z x dipole
side view (xy-plane) side view (yz-plane) top view (xz-plane)

y y z
directed
x z x antenna

side view (xy-plane) side view (yz-plane) top view (xz-plane)

Giuseppe Bianchi

Base stations in cellular

system
Often sectorized antennas used

z
z

x
sectorized
x antenna

top view, 3 sector top view, 6 sector

Giuseppe Bianchi

5
 Three means of propagation

ÎGround wave propagation

Îsky wave propagation
ÖIonospheric
ÎLine of Sight propagation

Giuseppe Bianchi

Ground Wave Propagation

Î0-2 MHz (e.g. AM radio)

Îfollows contour of the earth (no need for LOS)
Ö Bending by: current induced on earth
Ö Bending by: atmospheric diffraction
ÎVery long distances
Ö But signal dies off rapidly: need much power
Îreflection, refraction and scattering by objects on the ground
Giuseppe Bianchi

6
 Ionospheric or Sky Wave
Propagation

Î 2-30 MHz (HF)

Î Signal reflected from ionized layer of atmosphere back
down to earth (actually, rifracted)
Î Signal can travel a number of hops, back and forth
between ionosphere and earth’s surface
Ö By repeated reflection, communication can be established over 1000s of km
Î Examples: amateur radio, CB radio, International Broadcasting
Giuseppe Bianchi

Line-of-Sight Propagation

ÎTransmitting and receiving antennas

must be within line of sight
ÖSatellite communication – signal above 30 MHz not reflected
by ionosphere
ÖGround communication – antennas within effective line of
site (radio horizon) due to refraction (bending of wave in the
lower atmosphere)
Giuseppe Bianchi

7
 Radio horizon

Optical
horizon Radio
horizon
optical horizon : d = 3.57 h
radio horizon : d = 3.57 K ⋅ h K ≈ 4/3
d = [ Km], h = [m]

Giuseppe Bianchi

Antennas height reqs

(examples)
ÎAntenna = 100m, K=4/3
ÖD = 41.2 Km (radio horizon)

ÎReceiver antenna = 10m

(
d = 3.57 K ⋅ ht + K ⋅ hr )
ÖD = 54.26

Îhr = 10m, d=41.2, ht=?

Öht = 46.75

Giuseppe Bianchi

8
 Propagation impairments

ÎLine of sight

ÎReflection

ÎShadowing

Giuseppe Bianchi

Propagation impairments

BS
ÎDiffraction MS
Î When the surface
encountered has sharp edges
Æ bending the wave

BS
ÎScattering
Î When the wave encounters
objects smaller than the
wavelength (vegetation,
clouds, street signs)

Giuseppe Bianchi

9
 Signal attenuation
Signal power

Distance BS Æ MS
Giuseppe Bianchi

Multipath Characteristics
(not just attenuation)

ÎA signal may arrive at a receiver

Ömany different times
ÖFrom many different directions
Îdue to vector addition, signal may
ÖReinforce
ÖCancel
Îsignal strength differs
Öfrom place to place
Öfrom time to time!
Î(slow/fast fading)
Giuseppe Bianchi

10
 Slow fading – fast fading
Signal power

Distance BS Æ MS (m)

Distance BS Æ MS (km)
Giuseppe Bianchi

Radio Signal Fading

Signal strength (dB)

Short term fading

Long term fading

T

Time

Giuseppe Bianchi

11
 PART 1
Propagation Characteristics of
Wireless Channels

Lecture 1.2
Attenuation models

Giuseppe Bianchi

Power units - decibel

ÎDecibel (dB): logarithmic unit of intensity used

to indicated power lost or gained between two
signals
ÎNamed after Alexander Graham Bell.

10 log(P1 / P2 )
PA = 1 Watt
PB = 50 milliWatt
Æ PA = 13 dB greater than PB

Giuseppe Bianchi

12
 Decibels - dBm
ÎdBm = absolute value (reference= 1mW)
ÖPower in dBm = 10 log(power/1mW)
ÖPower in dBW = 10 log(power/1W)
» Not much used by us 1dBW=30dBm
ÎExamples
Ö10 mW = 10 log10(0.01/0.001) = 10 dBm
Ö10 µW = 10 log10(0.00001/0.001) = -20 dBm
Ö26 dBm = ___ 2W= ___ dBm?
ÖS/N ratio = -3dB Æ S = ___ X N?
ÎProperties & conversions
ÖdBm = 10 log10(P (W) / 1 mW) = P (dB) + 30 dBm
ÖP1 * P2 (dBm) = P1 (dBm) + P2 (dB)
P1 * P2 (dBm) = 10 log10(P1*P2 (W)/0.001) =
10log10(P1/0.001) + 10 log10P2 = P1 (dBm) + P2 (dB)
Giuseppe Bianchi

Computation with dB
ÎTransmit power
ÖMeasured in dBm
ÆEs. 33 dBm
ÎReceive Power
ÖMeasured in dBm
ÆEs. –10 dBm
ÎPath Loss
ÖReceive power / transmit power
ÖMeasured in dB
ÖLoss (dB) = transmit (dBm) – receive (dBm)
ÆEs. 43 dB = attenuation by factor 20.000

Giuseppe Bianchi

13
 Attenuation model for LOS
ÎDirect path between transmitter and
receiver
Öunobstructed Line-Of-Sight (LOS)
ÖRadio signal behaves like light in free space (straight line)
ÎReceive power:
ÖIn absence of obstacles, received power follows inverse
square law
Pr (d ) ∝ d −2
Æ(d = distance between sender and receiver)

Giuseppe Bianchi

free space model – ideal antennas

ÎIsotropic (omnidirectional) r
tx antenna in free space
ÖTransmitted power: Pt
ÖPower attenuation Pa at distance d:
down with sphere superficies Pt
Pa (d ) =
4πd 2
ÎPower received by isotropic
rx antenna
ÖPlanar wave Pr (d ) = Pa (d ) Ae
ÖAe = Effective Area
λ2
Ae =
4π
Giuseppe Bianchi

14
 free space model – real antennas
r
ÎNon isotropic tx antenna
ÖAntenna gain Gt
Gt Pt
Pa (d ) =
4πd 2

ÎNon isotropic rx antenna

ÖAntenna gain Gr
Pt Gt λ2
Pr (d ) = Pa (d )Gr Ae = Gr
4πd 2 4π

Giuseppe Bianchi

Friis Free-Space Model

summarizing all previous considerations
2
Pt Gt Gr λ2 Gt Gr ⎛ c ⎞
Pr (d ) = = P ⎜ ⎟ d >0
(4π ) 2 d 2 L L ⎜⎝ 4πfd ⎟⎠
t

Î Pt = transmitter power
Ö (W or mW)
Î Gt = transmitter antenna gain
Î Gr = transmitter antenna gain
Ö (dimensionless)
Î λ = c/f = RF wavelength (m)
Ö c = speed of light (3x108 m/s)
Ö f = RF frequency (Hz)
Î Pt Gt = Equivalent Isotropic
Radiated Power (EIRP)
Î L = other system losses (hardware)
Ö L >=1 (dimensionless)
Î d = distance between transmitter
and receiver (m)

Pr (d ) [dBm]= Pt [dBm] + 10 log10 Gt + 10 log10 Gr +

20 log10 λ − 20 log10 (4π ) − 20 log10 d − 10 log10 L
Giuseppe Bianchi

15
 Example
normalized
frequency [MHz] 900 900000000
speed of light [Km 300000 300000000
lambda (m) 0,333333333
gain Tx 1
Gain Rx 1
Loss 1
Ptx [W] 5
distance (Km) Prx W Prx dBm
200 8,80E-08 -40,56
400 2,20E-08 -46,58
600 9,77E-09 -50,10 -30,00
800 5,50E-09 -52,60
1000 3,52E-09 -54,54

received power (dBm)

1200 2,44E-09 -56,12 -40,00
1400 1,79E-09 -57,46
1600 1,37E-09 -58,62
1800 1,09E-09 -59,64 -50,00
2000 8,80E-10 -60,56
2200 7,27E-10 -61,39
2400 6,11E-10 -62,14 -60,00
2600 5,20E-10 -62,84
2800 4,49E-10 -63,48
3000 3,91E-10 -64,08 -70,00
3200 3,44E-10 -64,64 0 1000 2000 3000 4000 5000
3400 3,04E-10 -65,17
3600 2,71E-10 -65,66 distance (m)
3800 2,44E-10 -66,13
4000 2,20E-10 -66,58
4200 1,99E-10 -67,00
4400 1,82E-10 -67,41
4600 1,66E-10 -67,79
4800 1,53E-10 -68,16
5000 1,41E-10 -68,52

Giuseppe Bianchi

Path Loss (propagation loss)

positive value in dB

P ⎧⎪ L ⎛ 4πd ⎞
2
⎫⎪
PL(d )[dB ] = 10 log10 t = 10 log10 ⎨ ⎜ ⎟ ⎬=
Pr ⎪⎩ Gt Gr ⎝ λ ⎠ ⎪⎭
Gt Gr λ
= 20 log10 d − 10 log10 − 20 log10 =
L 4π
GG c
= 20 log10 d + 20 log10 f − 10 log10 t r − 20 log10 =
L 4π
GG
= 20 log10 d + 20 log10 f − 10 log10 t r − 147.56
L

Giuseppe Bianchi

16
 Free space loss
same as path loss, but part due to attenuation
in free space only (in dB)

−2
⎛ λ ⎞
L free (d ) = ⎜ ⎟
⎝ 4πd ⎠

⎡ λ ⎤ ⎡c / f ⎤
L free (d ) [dB ]= −20 log ⎢ = −20 log
⎣ 4πd ⎥⎦ ⎢⎣ 4πd ⎥⎦
= 20 log10 d + 20 log10 f − 147.56

Giuseppe Bianchi

Reference distance
Î If known received power at a reference distance do
from tx 2
Ö can calculate Pr(d) for any d ⎛ do ⎞
Pr (d ) = Pr (d o )⎜ ⎟
⎝d ⎠
⎛d ⎞ ⎛d ⎞
Pr (d )(dBm) = 10 log10 Pr (d o ) + 20 log10 ⎜ o ⎟ = Pr (d o )(dBm) + 20 log10 ⎜ o ⎟
⎝d ⎠ ⎝d ⎠

Î Must be smaller than typical distances encountered in

wireless communication systems;
Î Must fall in the far-field region of the antenna
Ö So that losses beyond this point are purely distance-dependent
Î Typical d0 selection: 100-1000m

Giuseppe Bianchi

17
 Example

Îd0 = 100m, for previous example

Gt Gr λ
Pr (d 0 )(dBm) = Pt (dBm) + 10 log10 + 20 log10 =
L 4πd
1
= 30 + 10 log10 5 + 0 + 20 log10 = −34,5 dBm
1200π
ÎPr(1000m)
d0
Pr (1000m) = Pr (d 0 ) + 20 log10
=
d
= −34,5 dBm − 20dB = −54,5dBm
Giuseppe Bianchi

Reference distance + frequency

ÎSimilarly, one may evaluate

received power at a reference
distance do from tx and a reference
frequency f0
Pt Gt Gr λ2 Pt Gt Gr (c / f 0 )
2
2
⎛fd ⎞
Pr (d , f ) = = ⋅ ⎜⎜ 0 0 ⎟⎟ =
(4π ) 2 d 2 L (4π ) 2 d 02 L ⎝ fd ⎠
= Pr (d 0 , f 0 )[dBm] − 20 log10 ( f ) − 20 log10 (d )
[where f and d are measured in multiple of
- f0 (typically 1 MHz)
- d0 (typically 1 m or 1Km)
Giuseppe Bianchi

18
 More realistic propagation
models
ÎInverse square power law
ÖWay too optimistic (ideal case)
ÖReal world: η-th power law

Pr (d ) ∝ d −η
Öwith η ranging up to as much as η=7
ÆIf tough environment (e.g., lots of foliage),
Ötypical values:
Æη=2 for small distances (20 dB/decade)
Æη=3 to η=4 (40 dB/decade) for mobile telephone distances
Öη higher in cities and urban areas; η lower in suburban or
rural areas.

Giuseppe Bianchi

Extended formulae
⎛d ⎞
Pr (d )(dB ) = 10 log10 Pr (d o ) + 10η log10 ⎜ o ⎟
⎝d ⎠
d_ref 1 Km
P_ref -51,5266 dBm (Ptx=10W; 900 MHz; 1000m)

distance prx (eta=2)prx (eta=3,5) prx (eta=4)

1 -51,5266 -51,5266206 -51,5266
1,2 -53,1102 -54,2979642 -54,6939
1,4 -54,4492 -56,6411018 -57,3717 -50
1,6 -55,609 -58,67082 -59,6914
1,8 -56,6321 -60,4611582 -61,7375 -55
received power (dBm)

2 -57,5472 -62,0626704 -63,5678

2,2 -58,3751 -63,5114144 -65,2235 -60
2,4 -59,1308 -64,834014 -66,7351
2,6 -59,8261 -66,0506877 -68,1256 -65
2,8 -60,4698 -67,1771517 -69,4129
3 -61,069 -68,2258645 -70,6115 -70 η=2
3,2 -61,6296 -69,2068698 -71,7326 η=3,5
3,4 -62,1562 -70,1283827 -72,7858 -75 η=4
3,6 -62,6527 -70,9972081 -73,7787
3,8 -63,1223 -71,8190464 -74,718 -80
4 -63,5678 -72,5987203 -75,609
4,2 -63,9916 -73,3403457 -76,4566
-85
4,4 -64,3957 -74,0474642 -77,2647
4,6 -64,7818 -74,7231447 -78,0369
1 2 3 4 5
4,8 -65,1514 -75,3700639 -78,7763 distance (Km)
5 -65,506 -75,9905707 -79,4854

Giuseppe Bianchi

19
 Realistic scenarios
Î obstructions between the transmitter and
receiver
Ö reflection, diffraction, scattering
Ö Propagation strongly influenced by environment (building characteristics,
vegetation density, terrain variation)
Ö Perfect conductors reflect waves; nonconductors absorb some energy
Î wave traverses multiple paths
Ö Radio waves arrive at receiver from different directions and with different
time delays
Î Resultant signal at receiving antenna is vector
addition of incoming signals
Ö signals can add constructively (resultant signal has large power) or
destructively (resultant signal has small power) depending on relative
phases

Software tools needed to analyze complex specific scenarios (ray-tracing)

Giuseppe Bianchi

Example scenarios:
LOS path non necessarily existing (and unique)
Example: city with large buildings;
No LINE OF SIGHT;
Diffraction; reflection

diffraction

reflection

Giuseppe Bianchi

20
 Example scenarios
LINE OF SIGHT +
Diffraction, reflection, scattering

LOS

Giuseppe Bianchi

Two-Ray Ground
Propagation Model
ÎTheoretical foundation for η=4
ÖTwo-ray model assumes one direct LOS path and one
reflection path reach receiver with significant power
ÖEasy to solve

Line-Of-Sight ray
ht
hr
reflected ray

Transmit and receive antennas at different height (in general)

Giuseppe Bianchi

21
 Two-ray model – geometry
d >> ht , hr
ddirect
ht
hr
θ( )θ

dreflect

1/ 2
⎧⎪ ⎛ h − h ⎞ 2 ⎫⎪ ⎧⎪ 1 ⎛ h − h ⎞ 2 ⎫⎪
d direct = d + (ht − hr ) = d ⎨1 + ⎜ t r ⎟ ⎬ ≈ d ⎨1 + ⎜ t r ⎟ ⎬
2 2

⎪⎩ ⎝ d ⎠ ⎪⎭ ⎪⎩ 2 ⎝ d ⎠ ⎪⎭
1/ 2
⎧⎪ ⎛ ht + hr ⎞ 2 ⎫⎪ ⎧⎪ 1 ⎛ ht + hr ⎞ 2 ⎫⎪
d reflect = d + (ht + hr ) = d ⎨1 + ⎜ ⎟ ⎬ ≈ d ⎨1 + ⎜ ⎟ ⎬
2 2

⎪⎩ ⎝ d ⎠ ⎪⎭ ⎪⎩ 2 ⎝ d ⎠ ⎪⎭
Giuseppe Bianchi

Two ray model – path analysis

⎧⎪ 1 ⎛ ht + hr ⎞ 2 ⎫⎪ ⎧⎪ 1 ⎛ ht − hr ⎞ 2 ⎫⎪ hh
d reflect − d direct ≈ d ⎨1 + ⎜ ⎟ ⎬ − d ⎨1 + ⎜ ⎟ ⎬=2 t r
⎪⎩ 2 ⎝ d ⎠ ⎪⎭ ⎪⎩ 2 ⎝ d ⎠ ⎪⎭ d

ÎEM waves travel for different distance

ÖSum up with different phase!
⎡ ⎛ d ⎞⎤
direct ray ∝ A cos ⎢2πf ⎜ t − direct ⎟⎥
⎣ ⎝ c ⎠⎦
⎡ ⎛ d reflect ⎞⎤
reflect ray ∝ B cos ⎢2πf ⎜⎜ t − ⎟⎥
⎣ ⎝ c ⎟⎠⎦
Ö A = attenuation along direct path
Ö B = attenuation along reflected path (reflection not ideal, in general)

Giuseppe Bianchi

22
Two ray model – field strength
ÎPhase difference
∆d 2π 4πht hr
∆ϕ = 2πf = ∆d =
c λ λd
ÎReceived field strength
ÖLet Edirect be the field strength given by direct ray.
ÖThen E = E [
1 + ρe − j∆ϕ
direct ]
ÎAssume ideal reflection (ρ=-1)
[ ]
E = Edirect 1 − e − j∆ϕ = Edirect [1 − cos ∆ϕ + j sin ∆ϕ ]
[
E = Edirect 1 + cos 2 ∆ϕ − 2 cos ∆ϕ + sin 2 ∆ϕ ]
1/ 2
=
1 − cos ∆ϕ ∆ϕ
= 2 Edirect = 2 Edirect sin
2 2
Giuseppe Bianchi

Two ray model – power

computation
ÎReceived power
ÖProportional to |E|2

Pr ∝ 4 Edirect sin 2 (∆ϕ / 2 )

PG G ⎛ λ ⎞ 2 ⎛ 2πht hr ⎞
2

Pr (d ) = t t r ⋅ ⎜ ⎟ ⋅ 4 sin ⎜ ⎟
L ⎝ 4πd ⎠ ⎝ λd ⎠

Giuseppe Bianchi

23
 Two ray model - conclusion
Î Typical values:
Ö ht ~ few tens of m 2πht hr ⎛ 2πht hr ⎞ ⎛ 2πht hr ⎞
2

Ö hr ~ couple of meters ≈ small ⇒ sin ⎜ 2

⎟≈⎜ ⎟
λd ⎝ λd ⎠ ⎝ λd ⎠
Ö λ ~ few tens of cm
Ö d ~ hundred meters – few km

Pt Gt Gr ⎛ λ ⎞ ⎛ 2πht hr ⎞
2 2
Pt Gt Gr ht2 hr2
Pr (d ) ≈ ⋅⎜ ⎟ ⋅ 4⎜ ⎟ = ⋅ 4
L ⎝ 4πd ⎠ ⎝ λd ⎠ L d

i.e. attenuation follows a 40 dB/decade rule!

Versus 20 dB/decade of the free-space model
Pr (d ) ∝ d −4

Giuseppe Bianchi

Design notes

ÎTypical assumptions for initial

system developement
Öη=2 power law attenuation for small distances
ÆFree space model
Öη=4 power law attenuation for large distances
ÆLOS + reflected ray model
ÖCross-over distance when d >> ht+hr
Æ(e.g., d > 10(ht + hr))

Giuseppe Bianchi

24
 Empirical models
ÎConsider specific scenarios
ÖUrban area (large-medium-small city), rural area
ÖModels generated by combining most likely ray traces
(LOS, reflected, diffracted, scattered)
ÖBased on large amount of empirical measurements
ÎAccount for parameters
ÖFrequency; antenna heights; distance
ÎAccount for correction factors
Ö(diffraction due to mountains, lakes, road shapes, hills, etc)
First model: Okumura, 1968
VERY complex due to many specific correction factors!
Giuseppe Bianchi

Okumura-Hata model
ÎHata (1980): very simple model to fit
Okumura results
ÎProvide formulas to evaluate path loss
versus distance for various scenarios
ÖLarge cities; Small and medium cities; Rural areas
ÖLimit: d>=1km

Parameters:
Æf = carrier frequency (MHz) Effective BS
Antenna height
Æd = distance BS Æ MS (Km)
Æhbs = (effective) heigh of base
station antenna (m)
Æhms = height of mobile antenna (m)
Giuseppe Bianchi

25
 Okumura-Hata: urban area
L path (dB ) = 69.55 + 26.16 log10 f +
+ (44.9 − 6.55 log10 hbs ) log10 d +
− 13.82 log10 hbs − a(hms )
Î a(hms) = correction factor to differentiate large from
medium-small cities;
Î depends on MS antenna height

large cities : a(hms ) = 3.2[log10 (11.75hms )] − 4.97 f ≥ 400 MHz

2

small - med cities : a(hms ) = [1.1 log10 f − 0.7]hms − [1.56 log10 f − 0.8]

Very small correction difference between large and small cities (about 1 dB)

Giuseppe Bianchi

Okumura-Hata: suburban &

rural areas
Î Start from path loss Lp computed for small and
medium cities

2
⎡ f ⎤
suburban : L path (dB ) = L p − 2 ⎢log10 ⎥ − 5.4
⎣ 28 ⎦
L path ( dB) = L p − 4.78[log10 f ] + 18.33 log10 f − 40.94
2
rural :

Giuseppe Bianchi

26
 Okumura-Hata: examples

150

140

130
path loss (dB)

120

110
large cities
100 small cities
suburbs
90
rural area
80
0 1 2 3 4 5 6 7 8 9 10
distance (km)

F=900MHz, hbs=80m, hms=3m

Giuseppe Bianchi

Okumura-Hata and η
Î Coefficient of Log(d) depends only on hbs
Î 10η = attenuation (dB) in a decade
Ö (d=1 Æ d=10)
Î The higher the BS, the lower the coefficient η
44
42
40
38
10η

36
34
32
30
0 20 40 60 80 100 120 140
base station height (m)

Giuseppe Bianchi

27
 Other empirical models
Î Lee’s model
Ö Use at 900MHZ
Ö For distances > 1km
Ö Based on measurements taken in three cities (including Philadelpia)
Ö More complex than Okumura-Hata
Î Walfish-Ikegami model
Ö For frequency range 800-2000 MHz
Ö Valid for microcellular distances (20m – 5 km)
Ö Adopted by European Cooperation in the field of Scientific and Technical
(COST) research as reference model for 3G systems
Î Indoor propagation models
Ö Include attenuation factors due to building penetration
Ö Account for number of walls, floors, reflection loss, etc
Ö Based on zones (large zone, middle zone, small zone, microzone)

Giuseppe Bianchi

PART 1
Propagation Characteristics of
Wireless Channels

Lecture 1.3
fading models

Giuseppe Bianchi

28
 Statistical nature of received
power
Signal strength (dB)

Short term fading

Mean value
predicted
by attenuation
model (constant
Long term fading at given )

Time (or movement)

Giuseppe Bianchi

Multipath: short term fading

Î Short-term fading
Ö Also: multipath fading
Ö Also: small-scale fading
Ö Also: fading
Î Generated by
superposition of same
signal travelling along
many paths BS
Î Received signal:

er (t ) = ∑k =1 ak cos(2πf 0t + φ k )
N

Phase depends on path length Multipath fading consequences

Î f0=carrier frequency
Î N=number of paths
Î ak,φk=amplitude & phase of
component k
Giuseppe Bianchi
ÖSmall movements of tx or rx (order of ½λ)
Æchange interference pattern
Ædrastic fluctuations in signal strength due to
constructive/destructive interference
Æ15-20 cm for 900 MHz

29
 Multipath analysis

er (t ) = ∑k =1 ak cos(2πf 0t + φ k ) =
N
recall that : cos(2πf 0t + φ k ) =
= cos(2πf 0t ) cos(φ k ) − sin (2πf 0t )sin (φ k )

= cos(2πf 0t )∑k =1 ak cos φ k − sin (2πf 0t )∑k =1 ak sin φk =

N N

= X cos(2πf 0t ) − Y sin (2πf 0t )

In the assumptions:
- N large (many paths)
- φk uniformly distributed in (0,2π)
- ak comparable (no privileged path such as LOS)
X,Y are gaussian, identically distributed random variables

Signal envelope: X 2 +Y 2 Rayleigh distribution

Giuseppe Bianchi

Rayleigh distribution
0,7

0,6 σ2 = variance of X and Y

probability distribution

0,5 Gaussian r.v.

sigma2=1
sigma2=2
f a (x) =
0,4
sigma2=4

= Pr ( x ≤ a < x + dx ) =
0,3

0,2
2
x − x 2σ 2
0,1
= 2e
0 σ
0 1 2 3 4 5
amplitude
∞
π
2

E [a ] = ∫ x ⋅
x −x
e 2σ 2
=σ = 1.253σ
0
σ 2
2
∞
π ⎛ π⎞
2

Var [a ] = ∫ x 2 ⋅
x −x
e 2σ 2
−σ 2 = σ 2 ⎜ 2 − ⎟ = 0.4292σ 2
0
σ 2
2 ⎝ 2⎠
Giuseppe Bianchi

30
 Signal power
amplitude : a = X 2 + Y 2 : rayleigh distribution
power : p = a 2 = X 2 + Y 2 : exponential distribution
ÎAverage power:
Ö 2σ2 (average over time)
ÎInstantaneous power
ÖRandom variable, with: −x
f p (x ) =
1 2σ 2
Æprobability density function: e
2σ x 2
−
Fp ( x ) = 1 − e 2σ
2
ÆProbability distribution function:

Giuseppe Bianchi

Outage probability
ÎProbability that received power is lower
than a given threshold
ÖBelow which signal cannot be correctly received

Î Average received power P0 (= 2σ2)

Î Minimum power threshold γ
−γ
outage probability : Pr( p ≤ γ ) = Fp (γ ) = 1 − e P0

Example 1: Example 2:
average power=100 µW; average power= -13 dBm;
lower threshold=15µW; lower threshold=-30 dBm;
Outage probability= 1-exp(-15/100) = 13,9% Outage probability= 1-exp(-1/50) = 1,98%

Giuseppe Bianchi

31
 Long-term fading

Î Signal multiply reflected and/or

scattered before taking multiple paths
to the receiver
Î movements large enough change
wave path
Î Result: long-term fading
Ö = lognormal fading
Ö = shadowing

Giuseppe Bianchi

Long-term fading statistics

lognormal distribution
⎛d ⎞
Pr (d )(dB) = 10 log10 Pr (d o ) + 10η log10 ⎜ o ⎟ + Y
⎝d ⎠
Î Y = 0 mean gaussian r.v.
with standard deviation σdB dB
Î Probability distribution (in dB):
− ( pdB − Pav )2
1 2σ dB
2
fY ( pdB ) = e
2π ⋅ σ dB

Giuseppe Bianchi

32
 Long-term fading and
attenuation plot
attenuation: η=4 after 100m; η=2 before 100m

-30

-40
no shadowing
-50 sigma=3 dB
received power (dBm)

sigma=8 dB
-60

-70

-80

-90

-100

-110

-120
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
distance (m)
Giuseppe Bianchi

Outage probability examples

average received power (dBm) -80
lognormal stdandard deviation (dB) 6

outage threshold outage prob

Excel computation:
-86 15,87% = DISTRIB.NORM (thr, med, stddev, VERO )
-88 9,12%
-90 4,78%
-92 2,28%
-94 0,98%
-96 0,38%
-98 0,13%
-100 0,04%

General case computation:

1) Convert to normal st. distr 2) Evaluate Prob[-g,g]=
thr − Pav =Erf(g)=-Erf(γ)
γ= <0
2σ dB Erf(g) (1-Erf(g))/2
3) Outage = (1-Erf(g))/2 =
g = −γ >0 = Erfc(g)/2
Giuseppe Bianchi

33
 normal distribution
0.4
σ=1

0.3

0.2
σ=2 Standard:
µ=0; σ=1
0.1
σ=3

-6 -4 -2 µ=0 2 4 6

− ( x − µ )2
1
f X ( x) = e 2σ 2

2π ⋅ σ
Giuseppe Bianchi

Cumulative normal distribution

1
µ=0; σ=1
0.8

0.6

0.4

0.2

-4 -2 2 4
x t2
−
Q( x ) = FX ( x) =
1
2π ∫e
−∞
2
dt

Giuseppe Bianchi

34
 erf - erfc
x
erf ( x )=
2
∫ dt
−t 2
e
π 0

erfc( x) = 1 − erf ( x)

t2
x x x −
2⋅(1 / 2 )
erf ( x )=
2 1 1
∫ e dt = ∫ e dt = ∫
2
−t 2
−t2

π 0 π −x (
2π ⋅ 1 / 2 ) −x
e dt

Normal distribution with µ=0, σ=1/sqrt(2)

Giuseppe Bianchi

Cumulative normal distribution and erfc

⎛ x ⎞ 1
erfc⎜ ⎟
⎝ 2 ⎠ 0.8
2
0.6

0.4

0.2

-4 -2 2 4

Giuseppe Bianchi

35
Normal distribution (standard)
-4 0,003%
-3,9 0,005%
-3,8 0,007%
-3,7 0,011%
100,000%
-3,6 0,016%
-3,5 0,023%
-3,4 0,034%
-3,3 0,048% 10,000%
-3,2 0,069%
-3,1 0,097%
-3 0,135%
-2,9 0,187% outage prob 1,000%
-2,8 0,256%
-2,7 0,347%
-2,6 0,466%
-2,5 0,621% 0,100%
-2,4 0,820%
-2,3 1,072%
-2,2 1,390%
-2,1 1,786% 0,010%
-2 2,275%
-1,9 2,872%
-1,8 3,593%
-1,7 4,457% 0,001%
-1,6 5,480%
-1,5 6,681% -4 -3 -2 -1 0
-1,4 8,076%
-1,3 9,680% n x sigma
-1,2 11,507%
-1,1 13,567%
-1 15,866%

Giuseppe Bianchi

PART 1
Propagation Characteristics of
Wireless Channels

Lecture 1.4
coverage area estimation
(cell sizing)

Giuseppe Bianchi

36
 Cell radius
ÎOpen issue, until now:
ÖHow do we determine the cell radius?

ÎSeems very simple: given

ÖPt = transmitted power (dBm)
ÖPth = threshold power (dBm)
ÆSensitivity of the receiver, i.e. minimum amount of received
power for acceptable performance
ÎPath loss computed as
ÖLp = Pt - Pth
ÎRadius computed from Lp
⎛d ⎞
ÖVia η-law propagation formula L p [d o ⇒ d ] = 10η log10 ⎜ o ⎟
⎝d ⎠
ÖVia Okumura-Hata formula (or other refined model)
Giuseppe Bianchi

Example (part 1)

ÎReceived power at 10 mt: 0.1 W

ÎThreshold power: Pth = -50 dBm
Îη = 3.7
Pr (10mt )[dBm ] = 10 log10 100 = 20
R
Pr ( R)[dBm ] = Pr (10mt )[dBm ] − 10η log10 = Pth
10
R R 70
20 − 37 log10 = −50 → log10 =
10 10 37
70
R = 10 ⋅10 37 = 780mt

Outage prob at cell border? 50%!!!

Giuseppe Bianchi

37
 Fading Margin
Î Previous computation does
not account for long-term
fading
Ö Need to keep it in count, as it does not
reduce when the MS makes small moves L p = Pt − Pth
Ö IDEA: reduce cell radius to account for
a “fading margin” M
L p = Pt − Pth − M

Î Fading Margin definition:

Ö M = average received power at cell border prob
(dB) – threshold power (dB)
ÆM=0 means that the power received at Mean path
M
loss
cell border is equal to the threshold
Power
ÆM=6 (dB) means that the power at cell
received at cell border is 4 x the power border
threshold
1% - 2%
Giuseppe Bianchi

Example (part 2)

ÎReceived power at 10 mt: 0.1 W

ÎThreshold power: Pth = -50 dBm
Îη = 3.7
ÎSlow fading, with σdB=4 dB
ÎIf we use a fading margin M=6 Æ
R
Pr ( R)[dBm ] = Pr (10mt )[dBm ] − 10η log10 = Pth + M
10
64
R
20 − 37 log10 = −50 + 6 → R = 10 ⋅10 37 = 537mt
10
What is the experienced outage at cell border?
Giuseppe Bianchi

38
 Outage probability (1)
Recall that slow fading has lognormal distribution. In dB:
− ( pdB ( r )− Pav ,dB ( r ))2

fY ( pdB (r ))=
1 2σ dB
2
e
2π ⋅ σ dB
pdB(r)= power (dBm) received at given distance r
Pav,dB(r)= mean power (dBm, from attenuation laws) at same distance

Outage occurs when received power < Pth

As Pth is given in dB:
Pout (r ) = ∫ fY ( pdB (r ))d pdB (r )=
Pth

−∞

1 ⎛ P (r ) − Pth ⎞
= erfc⎜⎜ av ,dB ⎟
2 ⎝ 2 ⋅ σ dB ⎟⎠

Giuseppe Bianchi

Outage probability (2)

Let’s rewrite Pout(r) in terms of power received at cell border R
and in term of the loss parameter η. Recall that:
η
⎛ R⎞
Pav (r ) = Pav (R ) ⋅ ⎜ ⎟ Pav ,dB (r ) = Pav ,dB (R ) − 10η log
r
⎯⎯→
⎝r⎠
dB
R
Hence, referring to Pav,dB(R)= P0(R)
⎛ P (R ) − Pth − 10η log(r / R ) ⎞
Pout (r ) = erfc⎜⎜ 0
1 From the definition of
⎟=
2 ⋅ σ ⎟ fading Margin M
⎝ 2 dB ⎠
1 ⎛ M − 10η log(r / R ) ⎞
= erfc⎜⎜ ⎟
⎟
2 ⎝ 2 ⋅ σ dB ⎠
Which represents the probability that an MS at generic distance r is subject
to outage, given that the cell has a fading margin M.
Giuseppe Bianchi

39
 Outage probability (3)
If outage probability is specificaly computed for a terminal placed
at cell border:

⎛ M ⎞
Pout (cell border ) = erfc⎜⎜
1
⎟
2 ⋅ σ ⎟
⎝ 2 dB ⎠

Giuseppe Bianchi

Example (part 3)
Î Received power at 10 mt: 30%

0.1 W
25%
Î Threshold power:
Pth = -50 dBm
outage probability

20%
Î η = 3.7
Î Slow fading, with 15%

σdB=4 dB
Î fading margin M=6 10%

5%
Î Cell size =537mt
Î Poutage= 6,68% 0%
400 450 500 550 600 650 700
terminal distance (mt)

⎛ 70 − 37 log(r / 10) ⎞
plot : Pout (r ) = erfc⎜⎜
1
⎟
2 ⋅ σ ⎟
⎝ 2 dB ⎠
Giuseppe Bianchi

40
 Outage probability area (1)
To compute the outage probability in a circular area with radius R, we just need
to integrate Pout(r) over the cell area:

Paout (R ) = Pout (r )2πr dr = Substituting and simplifying

1 R
πR 2 ∫0
R r ⎛ M − 10η log(r / R ) ⎞ dr
=∫ ⋅ erfc⎜⎜ ⎟ = Changing variable
⎟R
0 R ⎝ 2 ⋅ σ dB ⎠
1 ⎛ M − 10η log( x ) ⎞
= ∫ x ⋅ erfc⎜⎜ ⎟dx = Converting from dBs
⎟
0
⎝ 2 ⋅ σ dB ⎠
⎧ σ ln 10
⎪ σ = dB
1 ⎛ ln (m ) − η ln x ⎞ ⎪ 10
= ∫ x ⋅ erfc⎜ ⎟dx where ⎨
⎝ 2 ⋅σ ⎠ ⎪m = 10 10 = P0 (R )[ mW ]
0 M

⎪⎩ Pth [ mW ]
Giuseppe Bianchi

Outage probability area (2)

ln(m ) 2 ⋅σ
let Q1 = ; Q2 = ; then
2 ⋅σ η
⎛ ln x ⎞
Paout (R ) = ∫ x ⋅ erfc⎜⎜ Q1 −
1 ln x
⎟⎟dx = changing variable θ = Q1 −
Q2
0
⎝ Q2 ⎠
∞
= Q2 ∫ e 2Q2 (Q1 −θ ) ⋅ erfc(θ )dθ =
Q1
∞
⎡ 1 ⎤
= ⎢− e Q2 (2Q1 +Q2 )erf (θ + Q2 ) − e 2Q2 (Q1 −θ )erfc(θ )⎥ =
1
⎣ 2 2 ⎦ Q1

erfc(Q1 ) − e (2Q1Q2 +Q2 )erfc(Q1 + Q2 )

1 1 ∞ x

=
2
erfc ( x ) =
2 2
∫e ∫e
−t 2 −t 2
dt = 1 − dt
π π
2 2 x 0

D [erfc ( x )] = −
2
e−x
2

∫ erfc (x )dx = x ⋅ erfc (x ) −

1
e−x
2

π
Giuseppe Bianchi

41
 Outage probability area versus M
1,E+00

sigma = 4 dB

1,E-01
sigma = 6 dB Represents the fraction
sigma = 8 dB
of “not covered” area in
outage probability (area)

a cell.
1,E-02

η=4 in the example

1,E-03

Application example:
1,E-04
If σdB=6, and target
is 1% outage area,
1,E-05 must set M~10
0 5 10 15 20 25
fading margin M (dB)

NOTE : if CDF of normal standard available, but no erf - erfc available :

erfc(x )
1⎛ ⎛ x ⎞⎞
DISTRIB.NORM .ST ( x) = ⎜⎜1 + erf ⎜ ⎟ ⎟⎟ ⇒ = 1 − DISTRIB.NORM .ST ( 2x)
2⎝ ⎝ 2 ⎠⎠ 2

Giuseppe Bianchi

Outage probability at border

worst case computation
Outage probability for
1,E+00
user at cell border
sigma = 4 dB
sigma = 6 dB
1,E-01
Greater than Paout
outage probability (cell border)

sigma = 8 dB

(conservative)
1,E-02

Much simpler computation

⎛ M ⎞
Pout (R ) = erfc⎜⎜
1
1,E-03
⎟
⎟
2 ⎝ 2 ⋅ σ dB ⎠
1,E-04
η=4 in the example
1,E-05
0 5 10 15 20 25 example:
fading margin M (dB)
If σdB=6, and target
is 1% outage at border,
must set M~14

Giuseppe Bianchi

42
 Cell radius computation
Î Step 1:
Ö From outage probability target (es. not EXAMPLE
greater than 2%)
-Outage (border) = 5%
ÆOn a per area basis
ÆOr on a border cell basis
−σdB = 6dB
From SW (or tables)
Ö Plus radio channel information (η, σdB)
M=1.65 σdB = 9.87 dB
Ö Compute M
Î Step 2: -Acceptable performance: -90 dBm
Ö From Power threshold Pth, −η = 3.7
-Received power = -65 dBm @ 3 Km
Ö Plus transmitted power Pt
⎛d ⎞
Æor equivalent information (es. − 90 + m = −65 − 10η log10 ⎜⎜ ⎟⎟
received power at reference ⎝ d0 ⎠
distance) d~7.7 Km
Ö Compute radius

Giuseppe Bianchi

Normal distribution (standard)

-4 0,003%
-3,9 0,005%
-3,8 0,007%
-3,7 0,011%
100,000%
-3,6 0,016%
-3,5 0,023%
-3,4 0,034%
-3,3 0,048% 10,000%
-3,2 0,069%
-3,1 0,097%
-3 0,135%
outage prob

-2,9 0,187% 1,000%

-2,8 0,256%
-2,7 0,347%
-2,6 0,466%
-2,5 0,621% 0,100%
-2,4 0,820%
-2,3 1,072%
-2,2 1,390%
-2,1 1,786% 0,010%
-2 2,275%
-1,9 2,872%
-1,8 3,593%
-1,7 4,457% 0,001%
-1,6 5,480%
-1,5 6,681% -4 -3 -2 -1 0
-1,4 8,076%
-1,3 9,680% n x sigma
-1,2 11,507%
-1,1 13,567%
-1 15,866%

Giuseppe Bianchi

43
 Slow+Fast fading margins
power

Transmitted power

d0 d1 d2 USAGE: - determine cell radius

- determine Pt

Pth + M1 + M2
Short-term fading margin M2
Pth + M1
Long-term fading margin M1
Power threshold Pth = Pt - Ploss
1% - 2%

Used also in the concept of LINK BUDGET:

1% - 2% = Path Loss + M1 + M2
Giuseppe Bianchi

44
 PART 2
Cellular Coverage Concepts

Lecture 2.1
why cells

Giuseppe Bianchi

Coverage for a terrestrial zone

Signal OK if Prx > -X dBm
Prx = c Ptx d-4
greater Ptx Æ greater d

d
1 Base Station
N=12 channels
BS •(e.g. 1 channel = 1 frequency)

N=12 simultaneous calls

Giuseppe Bianchi

1
 Cellular coverage
target: cover the same area with a larger number of BSs

19 Base Station
12 frequencies
4 frequencies/cell

Worst case:
4 calls (all users in same cell)
Best case:
76 calls (4 users per cell)
Average case >> 12
Low transmit power
Key advantages:
•Increased capacity (freq. reuse)
•Decreased tx power
Giuseppe Bianchi

Cellular coverage (microcells)

many BS

Very low power!!

Unlimited capacity!!

Usage of same spectrum

(12 frequencies)
(4 freq/cell)

Disadvantage:
mobility management

Giuseppe Bianchi

2
 Cellular system architecture
Î 1 BS per cell Wired network

Ö Cell: Portion of territory

covered by one radio station MSC
MSC 11 MSC
MSC 22
Ö One or more carriers
(frequencies; channels) per
cell f5
f6
f7
f1
f4 f2
Î Mobile users full- f6 f3

duplex connected with

f5 f7 f6
f1 f5 f7
BS f4
f3
f2
f4
f1
f2
f6 f3
f5 f7
Î 1 MSC controls many f1

BSs f4
f3
f2

BS = Base Station
Î MSC connected to MSC = Mobile Switching Centre
PSTN PSTN = Public Switching Telephone Network
Giuseppe Bianchi

Cellular capacity
Î Increased via frequency reuse
Ö Frequency reuse depends on interference
Ö need to sufficiently separate cells
Æreuse pattern = cluster size (7 Æ 4 Æ 3):
discussed later

Î Cellular system capacity: depends on

Ö overall number of frequencies
ÆLarger spectrum occupation
Ö frequency reuse pattern
Ö Cell size
ÆSmaller cell (cell Æ microcell Æ picocell) = greater capacity
ÆSmaller cell = lower transmission power
ÆSmaller cell = increased handover management burden

Giuseppe Bianchi

3
 hexagonal cells
Î Hexagon:
A D Ö Good approximation for circle
B C B
A D A
C B C
B A D A
B C B
A D A D
C B C Ö Ideal coverage pattern
D A D Æno “holes”
B C Æno cell superposition

ÎExample case:
ÖReuse pattern = 4

Giuseppe Bianchi

Cells in real world

Shaped by terrain, shadowing, etc

Cell border: local threshold, beyond which neighboring BS signal
is received stronger than current one
Giuseppe Bianchi

4
 PART 2
Cellular Coverage Concepts

Lecture 2.2
Clusters and CCI

Giuseppe Bianchi

Reuse patterns D R

ÎReuse distance: 7
ÖKey concept 7 6 2
ÖIn the real world depends on 6 2
1
ÆTerritorial patterns (hills, etc) 1
5 3
5 3 4
ÆTransmitted power
4
» and other propagation issues
such as antenna directivity, Cluster: K = 7
height of transmission
antenna, etc D
ÎSimplified hexagonal 4
1
2
cells model: 3
1 1
Öreuse distance depends on 4 2 4 2
reuse pattern (cluster size) 3
1
3

ÖPossible clusters: 4 2
3
Æ3,4,7,9,12,13,16,19,…
K=4
Giuseppe Bianchi

5
 Reuse distance
Î General formula D = R 3K
Î Valid for hexagonal geometry
Î D = reuse distance
Î R = cell radius
Î q = D/R =frequency reuse factor

K q=D/R
3 3,00
4 3,46
7 4,58
9 5,20
12 6,00
13 6,24

Giuseppe Bianchi

Proof
v
Î Distance between two cell
centers:
Ö (u1,v1) ÅÆ (u2,v2)

(3,2) D= [(u 2 ] [
2
− u1 ) cos 30 o + (v2 − v1 ) + (u 2 − u1 ) sin 30 o ]
2

u Ö Simplifies to:
D = (u 2 − u1 ) 2 + (v2 − v1 ) 2 + (u 2 − u1 )(v2 − v1 )
(1,1)
Ö Distance of cell (i,j) from (0,0):
30°
D = i 2 + j 2 + ij 3 R
DR = i 2 + j 2 + ij
Ö Cluster: easy to see that
K = DR2 = i 2 + j 2 + ij
Ö hence:D = R 3K
Giuseppe Bianchi

6
 Clusters
Clusters:
• Number of BSs comprised in
a circle of diameter D
• Number of BSs whose inter-
distance is lower than D

K=12(i=3,j=1)

K=7(i=2,j=1) K=4 (i=2,j=0)

Giuseppe Bianchi

Possible clusters
all integer i,j values

i j K=ii+jj+ij q=D/R
1 0 1 1,73
1 1 3 3,00
2 0 4 3,46
2 1 7 4,58
2 2 12 6,00
3 0 9 5,20
3 1 13 6,24
3 2 19 7,55
3 3 27 9,00
4 0 16 6,93
4 1 21 7,94
4 2 28 9,17
4 3 37 10,54
4 4 48 12,00
5 0 25 8,66
5 1 31 9,64

Giuseppe Bianchi

7
 Co-Channel Interference
A
G E A Î Frequency reuse implies that
C F G E remote cells interfere with
D C F tagged one
A B D C
E A B Î Co-Channel Interference (CCI)
F G E A Ö sum of interference from
C F G remote cells
B D C F
A B D
S signal power (S)
E A =
N noise power (N S ) + interferin g signal power (I)
S signal power (S)
=
I interferin g signal power (I)
S S
≈ as N S small
N I
Giuseppe Bianchi

CCI Computation -
assumptions
Î Assumptions Î Key simplification
ÖNI=6 interfering cells Ö Signal for MS at distance R
ÆNI=6: first ring interferers Ö Signal from BS interferers at
only distance D
Æwe neglect second-ring
interferers
ÖNegligible Noise NS Dint R
Power R Power
ÆS/N ~ S/I Po D Po
Öd−η propagation law Dint ~ D
Æη=4 (in general)
ÖSame parameters for all BSs
ÆSame Ptx, antenna gains, etc

Giuseppe Bianchi

8
 CCI computation
S S cost ⋅ R −η By using the assumptions of
≈ = =
∑ −η same cost and same D:
NI
N I
k =1
cost ⋅ D
−η η
1 ⎛R⎞ 1 ⎛ D⎞ 1 η Results depend
= ⎜ ⎟ = ⎜ ⎟ = q on ratio q=D/R
NI ⎝ D⎠ NI ⎝R⎠ N I (q=frequency reuse factor)

Alternative expression: recalling that D = R 3K

η

(3K ) 2 = (3K )
−η
S S 1 ⎛ R ⎞ 1 η 2
≈ = ⎜ ⎟ =
N I N I ⎝ R 3K ⎠ NI 6
S (3K )
2
3
NI=6,µ=4 Æ = = K2
I 6 2
USAGE: Given an S/I target, cluster size K is obtained
Giuseppe Bianchi

Examples
Îtarget conditions: Îtarget conditions:
ÖS/I=9 dB ÖS= 18dB
Öη=4 Öη=4.2
ÎSolution: ÎSolution:

S
= 100.9 = 7.94 ≈ 8
S
[dB] = 5η log(3K ) − 10 log 6
I I
S (3K )
η
18 + 7.78
=
2
⇒K=
2 S
⋅ log(3K ) = = 1.23
I 6 3 I 21
η =4 101.23
K≥ = 5.63 ⇒ K = 7
K ≥ 2 .3 ⇒ K = 3 3

Giuseppe Bianchi

9
 S/I computation
assuming 6 interferers only (first ring)

K q=D/R S/I S/I dB

3 3,00 13,5 11,3
4 3,46 24,0 13,8
7 4,58 73,5 18,7
9 5,20 121,5 20,8
12 6,00 216,0 23,3
13 6,24 253,5 24,0
16 6,93 384,0 25,8
19 7,55 541,5 27,3
21 7,94 661,5 28,2
25 8,66 937,5 29,7

Giuseppe Bianchi

Additional interferers
B
A D Î case K=4
B C B Ö note that for each
A D A D cluster there are always
B C B C B NI=6 first-ring interferers
D A D A
C B C B C
A D A D
B C B C B
D A D A
C B C B C
A D A D
B C B C B
D A D A
B C B
D A In CCI computation, contribute of
B
additional interferers is marginal
Giuseppe Bianchi

10
 sectorization
Î Directional antennas

Î Cell divided into sectors Sector 3

f a , 2 L +1 L f a ,3 L
Î Each sector uses
different frequencies Sector 1
Ö To avoid interference at sector f a ,1 L f a , L
borders

Î PROS:
Sector 2
Ö CCI reduction
f a , L +1 L f a , 2 L
Î CONS:
Ö Increased handover rate CELL a
Ö Less effective “trunking” leads to
performnce impairments

Giuseppe Bianchi

CCI reduction via sectorization

three sectors case

ÎInferference from 2
cells, only A
G E A
ÖInstead of 6 cells C F G E
D C F
With usual approxs A B D C
(specifically, Dint ~ D) E A B
F G E A
⎡S ⎤ R −η ⎡S ⎤
⎢⎣ I ⎥⎦ o 2 D −η = 3 ⋅ ⎢⎣ I ⎥⎦
= C F G
120 omni B D C F
⎡S ⎤ ⎡S ⎤ A B D
⎢⎣ I ⎥⎦ o dB = ⎢⎣ I ⎥⎦ dB + 4.77 G E A
120 omni F
Conclusion: 3 sectors = 4.77 dB improvement
Giuseppe Bianchi

11
 6 sectors
Î60o Directional
antennas

ÎCCI reduction:
Ö1 interfereer only
Ö6 x S/I in the omni case
ÖImprovement: 7.78 dB

Giuseppe Bianchi

PART 2
Cellular Coverage Concepts

Lecture 2.3
teletraffic considerations,
teletraffic planning

Giuseppe Bianchi

12
 Traffic generated by one user
(statistical notion of traffic)
traffic)

1 user making phone calls Î How to characterize this process?

Ö statistical distribution of the
“BUSY” period
BUSY 1 Ö statistical distribution of the
“IDLE” period
Ö statistical characterization of the
process “memory”
IDLE 0
Æ E.g. at a given time, does the
TRAFFIC is a “stochastic process” time probability that a user starts a
call result different depending on
what happened in the past?
Î Traffic characterization suitable for traffic engineering
amount of busy time in ∆t
traffic intensity A i = lim =
∆t → ∞ ∆t
= (average number λ of calls per hour )× (average call duration τ ) =
= probability that, at a random time t, user is in BUSY state =
= mean process value
All equivalent (if stationary process)
Giuseppe Bianchi

Traffic generated by more than

one users
U1 Traffic intensity
(adimensional, measured in Erlangs):

U2 4
A = ∑ Ai = 4 Ai
i =1
U3

U4

⎛4⎞
P[k active calls] = ⎜⎜ ⎟⎟ Aik (1 − Ai )
4−k

⎝ ⎠
k
E [active calls] = 4 ⋅ Ai = A
TOT

Giuseppe Bianchi

13
 example
Î5 users
ÎEach user makes an average of 3 calls
per hour
ÎEach call, in average, lasts for 4 minutes
⎡ calls ⎤ 4
× [hours ] = [erl ]
1
Ai = 3⎢ ⎥
⎣ hour ⎦ 60 5

A = 5 × [erl ] = 1[erl ]
1
5 number of active users
0
probability
0,327680
Meaning: in average, there is 1 active call; 1 0,409600
but the actual number of active calls varies 2
3
0,204800
0,051200
from 0 (no active user) to 5 (all users active), 4 0,006400
with given probability 5 0,000320

Giuseppe Bianchi

Second example n. active users

0
1
binom
1
30
probab
1,3E-01
2,7E-01
cumulat
0,126213
0,396669
2 435 2,8E-01 0,676784

Î 30 users
3 4060 1,9E-01 0,863527
4 27405 9,0E-02 0,953564
Î Each user makes an 5
6
142506
593775
3,3E-02
1,0E-02
0,987006
0,996960
average of 1 calls per 7 2035800 2,4E-03 0,999397
hour 8
9
5852925
14307150
5,0E-04
8,7E-05
0,999898
0,999985
Î Each call, in average, 10 30045015 1,3E-05 0,999998

lasts for 4 minutes 11

12
54627300
86493225
1,7E-06
1,9E-07
1,000000
1,000000
13 119759850 1,9E-08 1,000000
⎛ 4⎞
A = 30 × ⎜1 ⋅ ⎟ = 2 Erlangs
14 145422675 1,7E-09 1,000000
15 155117520 1,3E-10 1,000000
⎝ 60 ⎠ 16
17
145422675
119759850
8,4E-12
5,0E-13
1,000000
1,000000
SOME NOTES: 18 86493225 2,6E-14 1,000000
19 54627300 1,2E-15 1,000000
-In average, 2 active calls (intensity A); 20 30045015 4,5E-17 1,000000
-Frequently, we find up to 4 or 5 calls; 21
22
14307150
5852925
1,5E-18
4,5E-20
1,000000
1,000000
-Prob(n.calls>8) = 0.01% 23 2035800 1,1E-21 1,000000
24 593775 2,3E-23 1,000000
-More than 11 calls only once over 1M 25 142506 4,0E-25 1,000000
26 27405 5,5E-27 1,000000
27 4060 5,8E-29 1,000000
TRAFFIC ENGINEERING: how many 28 435 4,4E-31 1,000000

channels to reserve for these users! 29

30
30
1
2,2E-33
5,2E-36
1,000000
1,000000
Giuseppe Bianchi

14
 A note on binomial coefficient computation
⎛ 60 ⎞ 60!
⎜⎜ ⎟⎟ = = 1.39936e + 12
⎝ ⎠
12 12!48!
but 60!= 8.32099e + 81 (overflow problems! ! )
⎛ 60 ⎞ ⎛ ⎛ 60 ⎞ ⎞
⎜⎜ ⎟⎟ = exp⎜⎜ log⎜⎜ ⎟⎟ ⎟⎟ = exp(log(60!) − log(12!) − log(48!)) =
⎝12 ⎠ ⎝ ⎝12 ⎠ ⎠
⎛ 60 12 48
⎞
= exp⎜ ∑ log(i ) − ∑ log(i ) − ∑ log(i )⎟ (no overflow! ! before exp...)
⎝ i =1 i =1 i =1 ⎠

⎛ 60 ⎞ 12
⎜⎜ ⎟⎟ Ai (1 − Ai ) =
48

⎝ ⎠
12
⎛ 60 12 48
⎞
= exp⎜ ∑ log(i ) − ∑ log(i ) − ∑ log(i ) + 12 log( Ai ) + 48 log(1 − Ai )⎟
⎝ i =1 i =1 i =1 ⎠
(no overflow! ! never! )

Giuseppe Bianchi

Infinite Users
Assume M users, generating an overall traffic intensity A
(i.e. each user generates traffic at intensity Ai =A/M).
We have just found that ⎛ A⎞
M

⎜1 − ⎟
⎛M ⎞
k
M! ⎛ A ⎞ ⎝ M⎠
P[k active calls, M users] = ⎜⎜ ⎟⎟ Aik (1 − Ai )
M −k
= ⎜ ⎟
⎝ ⎠
k ( M − k )!k! ⎝ M ⎠ ⎛ A⎞
k

⎜ 1 − ⎟
Let MÆinfinity, while maintaining the same overall traffic intensity A ⎝ M⎠

M −k
1 Ak ⎛ A⎞ ⎛ A⎞
P[k active calls, ∞ users] = lim
M!
⋅ ⋅ k ⋅ ⎜1 − ⎟ ⋅ ⎜1 − ⎟ =
M → ∞ (M − k )! k! M
⎝ M⎠ ⎝ M⎠
−A
⎡ − ⎤
M
M (M − 1)L (M − k + 1) ⎢⎛
−k
A k
A⎞ A⎥ ⎛ A⎞ Ak
= ⋅ lim ⋅ ⎜1 − ⎟ ⋅ ⎜1 − ⎟ = e− A
k! M → ∞ Mk ⎢⎝ M ⎠ ⎥ ⎝ M⎠ k!
⎣ ⎦

Giuseppe Bianchi

15
 Poisson Distribution
30%
poisson
A=2 erl
25% binomial (M=30)

20%
A=10 erl
15%

10%

5%

0%
0 2 4 6 8 10 12 14 16 18 20 22

Very good matching with Binomial

Ak
Pk ( A) = e −A (when M large with respect to A)

k! Much simpler to use than Binomial

(no annoying queueing theory complications)
Giuseppe Bianchi

Limited number of channels

THE most important problem U1
in circuit switching
Î The number of channels U2
C is less than the
number of users M
(eventually infinite) U3 X
Î Some offered calls will
be “blocked”
Î What is the blocking U4 X
probability?
Ö We have an expression for
P[k offered calls]
Ö We must find an expression for
P[k accepted calls]
Ö As: TOT

P[block] = P[C accepted calls] No. carried calls versus t

No. offered calls versus t
Giuseppe Bianchi

16
 Channel utilization probability
offered traffic: 2 erl - C=3

Î C channels available 35%

Î Assumptions: 30%

Ö Poisson distribution (infin. users) 25% offered calls

Ö Blocked calls cleared 20% accepted calls
Î It can be proven (from
15%
Queueing theory) that:
10%
P[k calls in the system, k ∈ (0, C)] =
P[k offered calls]
5%

= C 0%

∑ P[i offered calls]

0 1 2 3 4 5 6 7 8

i =0

(very simple result!)

P[C offered calls]
Î Hence: P[system full] = P[C accepted calls] = C

∑ P[i offered calls]

i =0
Giuseppe Bianchi

Blocking probability: Erlang-B

Î Fundamental formula for Î Efficient recursive computation
telephone networks available
Ao E1,C −1 ( Ao )
planning
Ö Ao=offered traffic in Erlangs E1,C ( Ao ) =
C + Ao E1,C −1 ( Ao )
C
A o 100,00%

Πblock = C C! j = E1,C ( Ao )
blocking probability

10,00%
Ao
∑j =0 j! 1,00%
C=1,2,3,4,5,6,7
0,10%

0,01%
0 1 2 3 4 5
offered load (erlangs)

Giuseppe Bianchi

17
 NOTE: finite users

Î Erlang-B obtained for the Î Erlang-B can be re-obtained

infinite users case as limit case
Î It is easy (from queueing Î MÆinfinity
theory) to obtain an Î AiÆ0
explicit blocking formula Î M·AiÆAo
for the finite users case:

Î ENGSET FORMULA: Î Erlang-B is a very good

approximation as long as:
⎛ M − 1⎞ Î A/M small (e.g. <0.2)
AiC ⎜⎜ ⎟⎟
= C ⎝ ⎠
C
Π block Î In any case, Erlang-B is a
⎛ M − 1⎞
∑
conservative formula
Aik ⎜⎜ ⎟⎟ Î yields higher blocking
k =0 ⎝ i ⎠ probability
Ao Î Good feature for planning
Ai =
M
Giuseppe Bianchi

Capacity planning
ÎTarget: support users with a given Grade
Of Service (GOS)
ÖGOS expressed in terms of upper-bound for the blocking
probability
ÆGOS example: subscribers should find a line available in the
99% of the cases, i.e. they should be blocked in no more than
1% of the attempts
ÎGiven:
ÆC channels
ÆOffered load Ao
ÆTarget GOS Btarget
ÖC obtained from numerical inversion of Btarget = E1,C ( Ao )

Giuseppe Bianchi

18
 Channel usage efficiency
Offered load (erl) Carried load (erl)

Ao C channels Ac = Ao (1 − B )

Ao B
Blocked traffic

Ac Ao (1 − E1,C ( Ao )) Ao
efficiency : η = = ≈ if small blocking
C C C
Fundamental property: for same GOS, efficiency increases as C grows!!
Giuseppe Bianchi

example
100,0%
A= 40 erl
A= 60 erl
A= 80 erl
blocking probability

A= 100 erl
10,0%

1,0%

0,1%
0 20 40 60 80 100 120
capacity C

GOS = 1% maximum blocking. 40 erl C >= 53 η = 74.9%

Resulting system dimensioning 60 erl C >= 75 η = 79.3%
and efficiency: 80 erl C >= 96 η = 82.6%
100 erl C >= 117 η = 84.6%
Giuseppe Bianchi

19
 Erlang B calculation - tables

Giuseppe Bianchi

Erlang B calculation - software

ÎErlang-B formula very easy to implement
ÖEven if some tricks needed for numerical accuracy
ÎErlang-B inversion not so easy
ÖSoftware tools

ÎOnline calculator:
Öhttp://mmc.et.tudelft.nl/~frits/Erlang.htm

ÖGiven two parameter, calculates the third

ÆN = number of circuits
ÆB = blocking probability
ÆA = offered load

Giuseppe Bianchi

20
 Application to cellular networks
Cell size (radius R) may be determined
on the basis of traffic considerations
Î First step: Î Third step:
Ö Given num channels and GOS Ö Given density of users
ÆC=50 available channels in a cell Æδ=500 users/km2
ÆBlocking probability<=2% Ö Evaluate cell radius
Ö Evaluate maximum cell (offered) load M M
δ= ⇒ R=
ÆFrom Erlang-B inversion(tables) πR 2
πδ
A=40.25 erl
Ö R~438m
Î Second step
Ö Given traffic generated by each user
ÆEach user: 4 calls/busy-hour
ÆEach call: 2 min in average
ÆAi=4x2/60=0.1333 erl/user
Ö Evaluate max num of users in cell
ÆM=301.87 ~ 302
Giuseppe Bianchi

Other example
Î Three service providers are planning to provide cellular service for an urban
area. The target GOS is 2% blocking. Users make 3 calls/busy-hour, each
lasting 3 minutes in average (Ai=3/20=0.15)
Ö Question: how many users can support each provider?
Î Provider A configuration: 20 cells, each with 40 channels
Î Provider B configuration: 30 cells, each with 30 channels
Î Provider C configuration: 40 cells, each with 20 channels
Î Provider A: Î Provider B: Î Provider C:
Ö 40 channels/cell Ö 30 channels/cell Ö 20 channels/cell
Ö at 2%: Ao=30.99 erl/cell Ö at 2%: Ao=21.93 erl/cell Ö at 2%: Ao=13.18 erl/cell
Ö 619.8 erl-total Ö 654.9 erl-total Ö 527.2 erl-total
Ö M=4132 overall users Ö M=4386 overall users Ö M=3515 overall users

Compare case A with C! The reason is the lower efficiency of 20 channels versus 40

Giuseppe Bianchi

21
 Sectorization and traffic
Î Assume cluster K=7
Î Omnidirectional antennas: CCI=18.7 dB
Î 120o sectors: CCI=23.4 dB
Î 60o sectors: CCI=26.4 dB

Î Sectorization yields to better CCI

Î BUT: the price to pay is a much lower trunking
efficiency!

Î With 60 channels/cell, GOS=1%,

Ö Omni: 60 channels Ao=1x46.95= 46.95 erl η=77.5%
Ö 120 :
o 60/3=20 channels Ao=3x12.03= 36.09erl η=59.5%
Ö 60o: 60/6=10 channels Ao=6x4.46= 26.76erl η=44.1%

Giuseppe Bianchi

conclusion
ÎThis module has given some hints regarding:
Ö Cell sizing via propagation considerations
Ö Frequency reuse via propagation considerations
Ö Cell planning via teletraffic consideration
Î Very elementary models
Ö But sufficient to understand what’s inside planning
ÎNo mobility!
Ö Teletraffic models need to be extended to manage handover rates!
Ö Blocking requirement for an handover call MUST be much lower
than blocking for a new incoming call
Æsevere math complications
ÆGuard channels for handover
ÆOut of the scopes of this class!

Giuseppe Bianchi

22
 PART 3
Introduction to GSM

Lecture 3.0
History

Giuseppe Bianchi

History of wireless
communication
Î 1896: Marconi
Ö first demonstration of wireless telegraphy
Ö tx of radio waves to a ship at sea 29 km away
Ö long wave transmission, high power req. (200 kW and +)
Î 1901: Marconi
Ö Telegraph across the atlantic ocean
Ö Close to 3000 Km hop!
Î 1907 Commercial transatlantic connections
Ö huge ground stations (30 by100m antennas)
Î 1915: Wireless telephony established
Ö NY – S. Francisco
Ö Virginia and Paris
Î 1920 Marconi:
Ö Discovery of short waves (< 100m)
Ö reflection at the ionosphere
Ö (cheaper) smaller sender and receiver, possible due to the invention of the
vacuum tube (1906, Lee DeForest and Robert von Lieben)

Giuseppe Bianchi

1
 History of wireless
communication
Î 1920's: Radio broadcasting became popular
Î 1928: many TV broadcast trials
Î 1930's: TV broadcasting deployment
Î 1946: First public mobile telephone service in US
Ö St. Louis, Missouri
Ö Single cell system
Î 1960's: Bell Labs developed cellular concept
Ö brought mobile telephony to masses
Î 1960’s: Communications satellites launched
Î Late 1970's: technology advances enable
affordable cellular telephony
Ö entering the modern cellular era
Î 1974-1978: First field Trial for Cellular System
Ö AMPS, Chicago

Giuseppe Bianchi

1st generation mobile systems

early deployment
ÎFirst system:
Ö NMT-450 (Nordic Mobile Telephone)
ÆScandinavian standard; adopted in most of Europe
Æ450 MHZ band
ÆFirst european system (Sweden, october 1981)
Î Italian history:
Ö 1966: first experiments (CSELT) at 160 MHZ
ÆRTMI (Radio Telefono Mobile Italiano)
ÆMarket: 1973
Ö First italian cellular system: 1985
ÆRTMS (Radio Telefono Mobile di Seconda Generazione)
Æ450 MHZ
Ö Evolution: 1990, TACS
ÆTotal Access Communication System
Æ900 MHZ

Giuseppe Bianchi

2
 1st generation mobile systems
ÎFirst generation: 1980’s ÎAnalog transmission
ÎSeveral competing standards in Ö Frequency modulation
different countries ÎVarious bands:
Ö NMT (Nordic Mobile Telephone) Ö NMT:
ÆScandinavian standard; adopted in most Æ450 MHz first
of Europe Æ900 MHz later
ÆFirst european system (Sweden, 1981) Ö TACS
Ö TACS (Total Access Communication Æ900 MHz
Systems), starts in 1985 Æ1230 bidirectional
ÆUK standard; A few of Europe, Asia, channels (25KHz)
Japan
Ö AMPS
Ö AMPS (Advanced Mobile Phone Æ800 MHz
Service)
ÆUS standard ÎToday still in use in low-
Ö C-Netz (Only in Germany) technology countries
Ö And not yet completely dismissed
Ö Radiocom 2000 (Only in France) in high-tech countries
Giuseppe Bianchi

2nd generation mobile systems

Î4 systems
ÖGlobal System for Mobile (GSM)
ÖDigital AMPS (D-AMPS), US
ÖCode Division Multiple Access
(IS-95) – Qualcomm,US
ÖPersonal Digital Cellular
(PDC),Japan
ÎGSM by far the
dominant one
ÖOriginally pan-european
ÖDeployed worldwide
Æ(slow only in US)
ÎBasic bands:
Ö900 MHz
Ö1800 MHz
Æ(Digital Cellular System:
DCS-1800)
Ö1900 MHz
Æ(Personal Communication
System:PCS-1900,US only)
ÎSpecifications for
ÖGSM-400 (large areas)
ÖGSM-800 (north america)

Giuseppe Bianchi

3
 Timing
Î1982: Start of GSM-specification in
Europe (1982-1990)
Î1983: Start of American AMPS
widespread deployment
Î1984 CT-1 standard (Europe) for
cordless telephones
Î1991 Specification of DECT
ÖDigital European Cordless Telephone (today: Digital
Enhanced Cordless Telecommunications)
- ~100-500m range, 120 duplex channels, 1.2Mbit/s data
transmission, voice encryption, authentication
Î1992: Start of GSM operation Europe-
wide
Î1994: DCS-1800
Giuseppe Bianchi

2 ½ generation mobile systems

GSM incremental extension
ÎHigh speed circuit switched data
(HSCSD)
ÆCircuit switched data communication
ÆUses up to 4 slots (1 slot = 9.6 or 14.4 Kbps)
ÎGeneral Packet Radio Service (GPRS)
ÆPacket data (use spectrum only when needed!)
ÆUp to 115 Kbps (8 slots)
ÎEnhanced Data-rates for Global Evolution
(EDGE)
ÆHigher data rate available on radio interface (3x)
» Up to 384 Kbps (8 slots)
» Thanks to new modulation scheme (8PSK)
» May coexist with old GMSK

Giuseppe Bianchi

4
3rd generation mobile systems
ÎUMTS (Universal Mobile
TelecommunicationSystem)
ÆITU standard: IMT-2000 (International Mobile
Telecommunication – 2000)
ÆUMTS forum created in 1996
ÆLater on 3GPP forum (bears most of standardization
activities)
ÖWideband CDMA radio interface
ÆBut several other proposals accepted as “compatible”
ÖRadio spectrum: 1885-2025 & 2110-2200 MHz
ÖAlready deployed in Japan
ÖTime to market in Italy: 2004?

Giuseppe Bianchi

Facts about wireless

communication
ÎWho has a cellular phone?
ÖUSA: Over 50% of US households
ÖItaly: from 2001, more wireless lines than wired
ÖWorld: from march 2002, 1 billion wireless cellular users
ÆMuch faster than projections!
ÆAugust 2000: 372 GSM networks, 362M customers
ÎRevenues:
Öglobal revenue from wireless portals predicted to grow from
$700M to $42 billion by 2005
ÖWLAN revenues predicted at $785M by 2004
ÖForecasting a 59 percent growth rate for wireless usage in
rural areas between 2000 and 2003

Giuseppe Bianchi

5
 PART 3
Introduction to GSM

Lecture 3.1
Architecture and components

Giuseppe Bianchi

GSM Network
high-
high-level view
MSC = Mobile Switching Center
= administrative region
PSTN
PSTN
Public
Publicswitched
switched
telephone
telephonenetwork
network
MSC MSC

Base Base
Station Station

PLMN
Public Land
Mobile Network

MSC role: telephone switching central with special mobility management capabilities
Giuseppe Bianchi

6
 GSM system hierarchy
MSC MSC region
LOCATION
BSC
AREA MSC: Mobile Switching Center
LA: Location Area
BSC: Base Station Controller
BTS BTS: Base Transceiver Station

Hierarchy: MSC region Æ n x Location Areas Æ m x BSC Æ k x BTS

Giuseppe Bianchi

GSM essential components

To fixed network OMC

(PSTN, ISDN, PDN)
GMSC
EIR AUC HLR VLR
MSC

BSC
BTS
MS Mobile Station
BTS BTS BTS Base Transceiver Station
BSC Base Station Controller
BTS BSC MSC Mobile Switching Center
GMSC Gateway MSC
OMC Operation and Maintenance Center
BTS
MS EIR Equipment Identity Register
AUC Authentication Center
HLR Home Location Register
VLR Visitor Location Register
Giuseppe Bianchi

7
 Î Two components:
GSM Sub-Systems Ö Fixed installed infrastructure
Æ The network in the proper sense
Ö Mobile subscribers
Æ MS: Mobile Station
Î Fixed infrastructure divided
operator into three sub-systems
Ö BSS: Base Station subsystem
Æ Manages transmission path from MS
to NSS
External Ö NSS: Network Switching Subsystem
Networks Æ Communication and interconnection
with other nets
OSS Ö OSS: Operational Subsystem
Æ GSM network administration tools

Users
NSS BSS MS

A Interface Radio Interface (Um)

Giuseppe Bianchi

PART 3
Introduction to GSM

Lecture 3.2
Mobile Station and addresses

Giuseppe Bianchi

8
 Mobile Station (MS)
GSM separates user mobility from equipment mobility,
by defining two distinct components

ÎMobile Equipment
ÆThe cellular telephone itself (or the vehicular telephone)
ÖAddress / identifier:
ÆIMEI (International Mobile Equipment Identity)

ÎSubscriber Identity Module (SIM)

ÆFixed installed chip (plug-in SIM) or
Æexchangeable card (SIM card)
ÖAddresses / identifiers:
ÆIMSI (International Mobile Subscriber Identity)
ÆMSISDN (Mobile Subscriber ISDN number)
» the telephone number!

Giuseppe Bianchi

Mobile Equipment structure

Terminal
Terminal Mobile
Mobile
Equipment
Equipment Termination
Termination Terminal Terminal Mobile
Equipment Adaptor Termination

Î Mobile Termination functions

Ö tRadio interface (tx, rx, signalling)
Î Terminal Equipment functions Mobile Equipment S Um
Ö User interface (microphone,
keyboard, speakers, etc); TE1 MT
Ö Functions specific of services
(telephony, fax, messaging, etc), R
independent of GSM
Î Terminal Adaptor functions TE2 TA MT
Ö Interfaces MT with different types
of terminals (PCs, Fax, etc.)
Giuseppe Bianchi

9
 Mobile Equipment Max Power
5 power classes

max power Type of

CLASS
(watt) terminal

I 20 vehicular

II 8 vehicular

III 5 portable
Normally used
IV 2 portable

V 0.8 portable

This was for 900 MHz – for 1800 MHz only two classes: 1W, and 0.25 W
Giuseppe Bianchi

IMEI
International Mobile Equipment Identity
Î Uniquely identifies the mobile equipment
Î 15 digits hierarchical address
Î assigned to ME during manifacturing and “type approval” testing
Ö Type approval procedure: guarantees that the MS meets a
minimum standard, regardless of the manifacturer
Î IMEI structure:

TAC – 6 digits FAC – 2 digits SNR – 6 digits SP – 1 digit

(Type Approval Code) (Final Assembly Code) (Serial Number) (Spare Digit)

centrally assigned assigned by assigned by Additional

upon type approval manufacturer manufacturer digit
available
Identifies place Unique for given
where ME was TAC+FAC
assembled or combination
manufactured
Giuseppe Bianchi

10
 IMEI management
Î Protection against stolen and malfunctioning terminals
Î Equipment Identity Register (EIR): 1 DataBase for each operator; keeps:
ÖWHITE LIST:
Ævalid IMEIs
ÆCorresponding MEs may be used in the GSM network
ÖBLACK LIST:
ÆIMEIs of all MEs that must be barred from using the GSM network
ÆException: emergency calls (to a set of emergency numbers)
ÆBlack list periodically exchanged among different operators
ÖGRAY LIST:
ÆIMEIs that correspond to MEs that can be used, but that, for some
reason (malfunctioning, obsolete SW, evaluation terminals, etc), need
to be tracked by the operator
ÆA call from a “gray” IMEI is reported to the operator personnel
Giuseppe Bianchi

SIM card
Subscriber Identity Module
Î Uniquely associated to a user
Ö Not to an equipment, as in first generation cellular networks
Î Stores user addresses
Ö IMSI
Ö MSISDN
Ö Temporary addresses for location, roaming, etc
Î authentication and encryption features
Ö All security features of GSM are stored in the SIM for maximum protection
Æsubscriber’s secret authentication key (Ki)
ÆAuthentication algorithm (“secret” algorithm - A3 – not unique)
ÆCipher key generation algorithm (A8)
Î Personalization
Ö SIM stores user profile (subscribed services)
Ö RAM available for SMS, short numbers, user’s directory, etc
Ö Protection codes
ÆPIN (Personal Identification Number, 4-8 digits)
ÆPUK (PIN Unblocking Key, 8 digits)
Giuseppe Bianchi

11
 IMSI
International Mobile Subscriber Identity
Î Uniquely identifies the user (SIM card)
Î GSM-specific address
Ö unlike MSISDN - normal phone number
Î 15 digits hierarchical address
Î assigned by operator to SIM card upon subscription
Î IMSI structure:

MCC – 3 digits MNC – 2 digits MSIN – max 10 digits

(Mobile Country Code) (Mobile Network Code) (Mobile Subscriber Identification Number)

Internationally standardized; Identifies operator Uniquely identifies subscriber

identifies operator country network (PLMN) in the operator network
within country
TIM=01 OMNI=10
Italy: 222
WIND=88 BLU=98
Giuseppe Bianchi

MSISDN
Mobile Subscriber ISDN Number
Î MSISDN: the “usual” telephone number
Ö Follows international ISDN numbering plan (ITU-T E.164 recommendations)
Ö Structure:
CC – up to 3 digits NDC – 3 digits (for PLMN) SN – max 10 digits
(Country Code) (National Destination Code) (Subscriber Number)

Î GSM is the first network to distinguish

Ö The user identity (i.e. IMSI)
Ö From the number to dial (i.e. MSISDN)
Î Separation IMSI-MSISDN protects confidentiality
Ö IMSI is the real user address: never public!
Ö Faking false identity: need signal IMSI to the network; but IMSI hard to obtain!
Î Separation IMSI-MSISDN allows
Ö Easy modification of numbering and routing plans
Î single IMSI may be associated to several MSISDN numbers
Ö E.g. different services (fax, voice, data, etc) may be associated with different
MSISDN numbers
Giuseppe Bianchi

12
 Temporary addresses
Î TMSI - Temporary Mobile Î MSRN - Mobile Station
Subscriber Identity Roaming Number
Ö 32 bits Ö An MSISDN number
Ö assigned by VLR within an ÆCC, NDC of the visited network
administrative area ÆSN assigned by VLR
Æhas significance only in this area Ö Used to route calls to a roaming MS
Ö transmitted on the radio interface ÆSubscriber Number (SN)
instead of IMSI assigned to provide routing
Æ reduces problem of information towards actually
“eavesdropping” responsible MSC

Giuseppe Bianchi

PART 3
Introduction to GSM

Lecture 3.3
Fixed Infrastructure

Giuseppe Bianchi

13
 Components and interfaces
Components
MS MS Mobile Station
BTS Base Transceiver Station
Um BSC Base Station Controller
MSC Mobile Switching Center
BTS OMC Operation and Maintenance Center
EIR Equipment Identity Register
Abis AUC Authentication Center
Other HLR Home Location Register
BSC VLR Visitor Location Register
MSC E BSS
A Interfaces
Other
MSC OMC Um Radio Interface
Networks
B Abis BTS-BSC
C A BSS-MSC
F B MSC-VLR
EIR HLR VLR C MSC-VLR
D D HLR-VLR
G E MSC-MSC
F MSC-EIR
Other
AUC G VLR-VLR
VLR

Giuseppe Bianchi

Base Station Sub-System

Um - Radio
Interface
BSS OSS
BTS

BTS BSC

A
BTS A-bis Interface
Interface

ÖBase Transceiver Station (BTS)

ÆTransmitter and receiver devices, voice coding & decoding, rate adaptation for data
ÆProvides signaling channels on the radio interface
ÆLimited signal and protocol processing (error protection coding, link layer LAPDm)
ÖBase Station Controller (BSC)
Æperforms most important radio interface management functions:
ÆRadio channels allocation and deallocation; handover management; …
Giuseppe Bianchi

14
 Base Transceiver Station - BTS
Um Interface
(to MS)

Abis Interface (to BSC)

Output HF

Transmission
filter Transmitter

Slow freq.
Hopping
TRX

System
Digital
Signal
Input HF Processing
Filter Receiver

Operation and Maintenance Functionality/clock distribution

TRX radio interface functions:

- GMSK modulation-demodulation
- channel coding In essence, BTS is
- encryption/decryption
- burst formatting, interleaving a complex modem!
- signal strength measurements
- interference measurements
Giuseppe Bianchi

BTS – maximum power

Î GSM 900 Î DCS 1800
Ö 320 W Ö 20 W
Ö 160 W Ö 10 W
Ö 80 W Ö5 W
Ö 40 W Ö 2.5 W
Ö 20 W Ö 1.6 W (micro-BTS)
Ö 10 W Ö 0.5 W (micro-BTS)
Ö5 W Ö 0.16 W (micro-BTS)
Ö 2.5 W
Ö 0.25 W (micro-BTS)
Ö 0.08 W (micro-BTS)
Ö 0.03 W (micro-BTS)

Giuseppe Bianchi

15
 Base Station Controller - BSC

DB contains
BTS-1 DB - state information for all BSS
- BTS software
BTS-2
X From/to MSC
switch FUNCTIONS:
Î switch calls from MSC to
matrix
correct BTS
Ö and conversely
BTS-K
Î Protocol and coding
conversion
Ö for traffic (voice) &
1 BSC may control signaling (GSM-specific
to ISDN-specific)
up to 40 BTS
Î Manage MS mobility
Î Enforce power control
Giuseppe Bianchi

Transcoding and Rate Adaptation

BTS:
-collects speech traffic
-Deciphers and removes error protection
-Result:
-13 kbps air-interface GSM speech-coded signal Transcoding and
MSC: Rate Adaptation
-A modified ISDN switch Unit (TRAU)
-Needs to receive ISDN-coded speech needed!
-64 kbps PCM format (A-law)

Rationale: re-use existing ISDN switches & protocols

Giuseppe Bianchi

16
 TRAU possible placements

13 kbit/s 64 kbit/s 64 kbit/s

On BTS BSC MSC
BTS TRAU

On BSC 13 kbit/s 16 kbit/s 64 kbit/s

BTS BSC TRAU MSC

64 kbit/s
13 kbit/s (4x16 sub-mux)
On MSC 16 kbit/s
BTS BSC TRAU MSC

Why 16 kbps instead of 13? Inband signalling needed for BTS control of TRAU
(TRAU needs to receive synchro & decoding information from BTS)
Giuseppe Bianchi

Network Switching Sub-System

ÎElements:
ÖMobile Switching Center (MSC) / Gateway MSC (GMSC)
ÖHome Location Register (HLR ) / Authentication Center (AuC)
ÖVisitor Location Register (VLR)
ÖEquipment Identity Register (EIR)
ÎFunctions:
ÖCall control
ÖUser management
ÎInter-component communication
ÖVia SS7 signalling network
ÖWith suitable extensions (e.g. MAP – Mobile Application Part)

Giuseppe Bianchi

17
 Mobile Switching Center - MSC
Î An ISDN switch (64 kbps channels)
Î Performs all the switching and routing functions
of a fixed network switching node
Î PLUS specific mobility-related functions:
Ö Allocation and administration of radio resources
Ö Management of mobile users
Æregistration, authentication
Æhandover execution and control
Æpaging

Î A PLMN (operator network) has,

in general, many MSC
Ö Each MSC is responsible of a set of BSS
Æ(note: a BSS refers to just 1 MSC, not many)

Giuseppe Bianchi

Home Location Register - HLR

Î 1 database per operator (PLMN)
Ö In principle; in practice may be
ÆN-plicated for reliability reasons
ÆIn large operator networks, there may be 2+ HLR with distinct
information, although MSISDN-HLR association needs to be
introduced (e.g. first two digits of the Subscriber Number)
Î HLR entries:
Ö Every user / MSISDN that has subscribed to the operator
Î Stores:
Î Permanent information associated Î Temporary information associated
to the user to the user
Ö IMSI, MSISDN Ö Link to current location of the user:
Ö Services subscribed ÆCurrent VLR address (if avail)
Ö Service restrictions (e.g. roaming ÆCurrent MSC address (if avail)
restrictions) ÆMSRN (if user outside PLMN)
Ö Parameters for additional services
Ö info about user equipment (IMEI)
Ö Authentication data
Giuseppe Bianchi

18
 Authentication Center - AUC
ÎAssociated to HLR
ÖEventually integrated with HLR
ÎSearch key: IMSI
ÎResponsible of storing security-relevant
subscriber data
ÖSubscriber’s secret key Ki (for authentication)
ÖEncryption key user on the radio channel (Kc)
ÖAlgorithms to compute volatile keys used during
authentication process

Giuseppe Bianchi

Needed, as fixed network

Gateway MSC – GMSC switches are not mobile
capable!!

X X GMSC task: query HLR for

current MS location

(if fixed network switches

X X X were able to query HLR,
direct connection with
local MSC would be available)
GMSC
MSC
MSC
MSC HLR

PLMN
Public Land
Mobile Network

Giuseppe Bianchi

19
 Visitor Location Register - VLR
Î At most 1 database per MSC
Ö Generally, joint MSC-VLR implementation
ÆNo need to carry heavy MSC-VRL signalling load on network links
Ö but 1 VLR may serve many MSCs
Î VLR entries:
Ö Every user / MSISDN actually staying in the administrative area of the
associated MSC
ÆEntry created when an MS enters the MSC area (registration)
Ö NOTE: may store data for roaming users (subscribed to different operators)
Î Stores:
Î Subscriber and subscription data Î Tracking and routing information
Ö IMSI, MSISDN Ö Mobile Station Roaming Number
Ö Parameters for additional services (MSRN)
Ö info about user equipment (IMEI) Ö Temporary Mobile Station Identity
Ö Authentication data (TMSI)
Ö Location Area Identity (LAI) where MS
has registered
ÆUsed for paging and call setup
Giuseppe Bianchi

Operation & Maintenance

Sub-system (OSS)
ÎNetwork measurement and control functions
ÎMonitored and initiated from the OMC
(Operation and Maintenance Center)
ÎBasic functions
ÖNetwork Administration
Æconfiguration, operation, performance management, statistics
collection and analysis, network maintenance
ÖCommercial operation & charging
ÆAccounting & billing
ÖSecurity Management
ÆE.g. Equipment Identity Register (EIR) management
O&M functions based on ITU-T TMN standards (Telecommunication
Network Management) – complex topic out of the scopes of this course
Giuseppe Bianchi

20
 PART 4
GSM – Radio Interface

Lecture 4.1
Physical channels

Giuseppe Bianchi

GSM Radio Spectrum

Frequency [MHz]
Î 2 x 25 Mhz band
960 Ö Duplex spacing: 45 MHz
Î 124 carriers x band
DOWNLINK
Ö 200 KHz channels
BS Æ MS
Ö Suggested use: only 122
ÆUse top & bottom as additional guard
935
Î 8 TDMA slots x carrier
Ö full rate calls – 13 Kbps
Ö If half-rate used, 16 calls at 6.5 kbps
915
890.4
1 2 3 4 5 6 7 8
UPLINK
MS Æ BS
Fuplink (n ) = [890.2 + 0.2(n − 1)] MHz
890.2

Fdwlink (n ) = [935.2 + 0.2(n − 1)] MHz

“guard band”
890
Giuseppe Bianchi

1
 Adjacent channels
(due to GMSK)

60dB 35dB

Specification: 9dB
In practice, due to power control and shadowing, adjacent channels
Cannot be used within the same cell…
Giuseppe Bianchi

Physical channel
Î 200 KHz bandwidth + GMSK modulation
Ö 1625/6 kbps gross channel rate (270.8333 kbps)
Î 1 time slot = 625/4 bits
Ö 156.25 bits
Ö 15/26 ms = 576.9 µs

time time
slot slot
0 7 time

577 µs

1 frame = 60/13 ms = 4.615 ms

26 frames = 120 ms (this is the key number)

Giuseppe Bianchi

2
 Hybrid FDMA-TDMA
physical channel = (time slot, frequency)
frequency Total n. of channels: 992

200 KHz
200 KHz
200 KHz
200 KHz
200 KHz
200 KHz slot
200 KHz
200 KHz
200 KHz
577us 577us 577us 577us 577us 577us 577us 577us
time
Giuseppe Bianchi

DCS 1800 radio spectrum

Î Greater bandwidth available
Ö EUROPE: 75 MHz band
Æ1710-1785 MHz uplink; 1805-1880 MHz downlink
Ö ITALY: 45 MHz band from 2005
Æ1740-1785 MHz uplink; 1835-1880 MHz downlink
Î Same GSM specification
Ö 200 KHz carriers
ÆA total of 374 carriers (versus124 in GSM)
Î DCS 1800 operators
Ö Common rule in most of the countries:
ÆFirst and second operators @ 900 MHz; Third etc @1800 MHz
ÆDCS 1800 deployment (1996+):
» 15 MHz (=75 carriers) to Wind; 7.5 (=37 carriers) to first and second
operator (plus existing 27 GSM 900 carriers)

Giuseppe Bianchi

3
 Other GSM bands
ÎExtended GSM (E-GSM) band
ÖUplink: 880-915 MHz
ÖDownlink: 925-960 MHz
ÎOther bands:
Ö450 MHz Æ (450.4-457.6 up; 460.4-467.6 down)
Ö480 MHz Æ (478.8-486 up; 488.8-496 down)
Ö1900 MHz Æ (1850-1910 up; 1930-1990 MHz)

Giuseppe Bianchi

Duplexing
- MS uses SAME slot number on uplink and downlink

- Uplink and downlink carriers always have a 45 MHz separation

-I.e. if uplink carrier is 894.2 Æ downlink is 919.2

-3 slot delay shift!!

0 1 2 3 4 5 6 7 DOWNLINK

UPLINK 0 1 2 3 4 5 6 7

MS: no need to transmit and receive in the same time

on two different frequencies!

Giuseppe Bianchi

4
 Slow Frequency hopping
(optional procedure within individual cell)
f7
f6
f5
f4
f3
f2
f1
Hopping sequence (example):
… Æ f1 Æ f2 Æ f5 Æ f6 Æ f3 Æ f7 Æ f4 Æ f1…

Slow = on a per-frame basis

- 1 hop per frame (4.615 ms) = 217 hops/second
Physical motivation:
- combat frequency-selective fading
- combat Co-Channel Interference
next slot may not interferere with adjacent cell slot (different hopping sequence)
- improvements: acceptable quality with 9 dB SNR versus 11 dB
Giuseppe Bianchi

Structure of a TDMA slot

Normal burst
Training
TB DATA S S Data TB GP
sequence
3 57 1 26 1 57 3 8.25

148 bit burst

156.25 bit (15/26 ms = 0.577 ms)

Î Symmetric structure
Î DATA: 2 x 57 data bits
Ö 114 data bits per burst
Ö “gross” bits (error-protected; channel coded)
Ö “gross” rate: 24 traffic burst every 26 frames (120 ms)
Æ22.8 kbps gross rate
Æ13 kbps net rate!
Î S: 2 x 1 stealing bit
Ö Also called stealing flags, toggle bits
Ö Needed to grab slot for FACCH (other signalling possible)

Giuseppe Bianchi

5
 Tail & training bits
Î 2 x TB = 3 tail bits set to 000
Ö At start and end of frame
Ö Leave time available for transmission power ramp-up/down
Ö Assures that Viterbi decoding starts and ends at known state

Î 26 bit training sequence

Ö Known bit pattern (8 Training Sequence Code available)
Ö for channel estimation and synchronization
Ö Why in the middle?
ÆBecause channel estimate reliable ONLY when the radio channel
“sounding” is taken!
ÆMultipath fading rapidly changes the channel impulse response…

Giuseppe Bianchi

Training sequences

Different codes used in adjacent cells! Avoids training sequence

Disruption because of co-channel interference.

Giuseppe Bianchi

6
 Power mask for Normal Burst

7.6 bits 4.9 bits

156.75 bits; 162.2 bits

Giuseppe Bianchi

Guard Period rationale

BTS
d

Î Assume the following

synchro mechanism:
Ö BTS transmits at time 0 BTS downlink tx 1 2 3 4
Ö MS receives at time d/c
Ö MS transmits at time 3+d/c MS downlink rx 1 2 3 4
Ö BTS receives at time 3+ 2d/c MS uplink rx 1

Î Offset depending on d! BTS uplink rx 1

1
Expected RX time!
Giuseppe Bianchi

7
 Guard period sizing
dwlink slot 1 … dwlink slot 4

dwlink slot 1 … dwlink slot 4

MS time BTS time

… uplink slot 1

… uplink slot 1

Maximum cell radius:

GTbits 2d c ⋅ GTbits 300000 ⋅ 8.25

= → d= ≈ ≈ 4.5 Km
Crate c 2Crate 2 ⋅ 270833
Is there something wrong? (GSM says that cells go up to 35 km)
Giuseppe Bianchi

Frame synchronization
TA (transmitted in the SACCH)
BTS

Î Timing Advance (TA)

Ö Parameter periodically transmitted by dwlink slot 4
BTS during MS activity
Ö 6 bits = 0-63
dwlink slot 4
Ö Meaning: anticipate transmission of
TA bits BTS time TA
MS time
Ö TA=0: no advance uplink slot 1
Æ I.e. transmit after 468.75 bits
after downlink slot
Ö TA=63: uplink slot 1
Æ Transmit after 405.75 bits time uplink slot 1
TA avoids collision!
Giuseppe Bianchi

8
 Timing Advance analysis
Î Downlink propagation delay:
Ö d/c
Î Uplink propagation delay:
Ö d/c
Î Uplink delay with TA:
Ö d/c-TA
Î Perfect resynchronization occurs when
Ö TA = 2d/c
Î Maximum cell size for perfect resync:

31.5 [bits ]
⋅ 300000 [km / s ] = 34.89 [km]
TA
d= ⋅c =
2 270833 [bits / s ]
8.25 bits Guard time additionally available for imperfect sync (+/- error)
Giuseppe Bianchi

And when the user is not connected?

But wants to connect…

Solution: USE A DIFFERENT BURST FORMAT

Access Burst: much longer Guard Period available
drawback: much less space for useful information

Access burst
Training
TB Data TB GP
sequence
8 41 36 3 68.25

88 bit burst
156.25 bit (0.577 ms)

No collision with subsequent slot for distances up to 37.8 km

Giuseppe Bianchi

9
 Other burst formats in GSM
Î5 different bursts available

ÖNormal Burst
ÖAccess Burst

ÖFrequency Correction Burst

ÖSynchronization Burst
ÖDummy Burst

Giuseppe Bianchi

Frequency Correction Burst

Frequency Correction Burst
TB Fixed bit pattern (all 0s) TB GP
3 142 3 8.25

ÎAll 0s burst (TB=0, too)

ÎAfter GMSK modulation:
ÖSine wave at freference+1625/24 kHz (67.7083 kHz)
ÎActs as a “beacon”
ÖWhen an MS is searching to detect the presence of a carrier
ÎAllows an MS to keep in sync with
reference frequency

Giuseppe Bianchi

10
 Dummy Burst
Dummy Burst
Training
TB Fixed bit pattern Fixed bit pattern TB GP
sequence
3 58 58 3 8.25
26

ÎUsed to fill inactive bursts on the BCCH

ÎGuarantees more power on BCCH than
that on other channels
ÖImportant, when MS needs to find BCCH

Giuseppe Bianchi

Synchronization Burst
Synchronization Burst
Training
TB Sync data Sync data TB GP
sequence
3 39 39 3 8.25
64

ÎLonger training sequence

ÖIt is the first burst an MS needs to demodulate!
Ö1 single training sequence
ÎData field:
ÖContains all the information to synchronize the frame
Æi.e. synchronize frame counter
ÖContains the BSIC (Base Station Identity Code, 6 bits)
Æ3 bits network code (operator)
» Important at international boundary, where same frequencies
can be shared by different operators
Æ3 bits color code
ÆTo avoid listening a signal from another cell, thinking it
comes from the actual one!
Giuseppe Bianchi

11
 PART 4
GSM – Radio Interface

Lecture 4.2
logical channels

Giuseppe Bianchi

Logical vs Physical channels

Logical channels
(traffic channels, signalling (=control) channels)

Physical channels
(FDMA/TDMA)

Î Physical channels
Ö Time slots @ given frequencies
Ö Issues: modulation, slot synchronization, multiple access techniques,
duplexing, frequency hopping, etc
Î Logical channels
Ö Built on top of phy channels
Ö Issue: which information is exchanged between MS and BSS

Giuseppe Bianchi

12
 Logical – physical mapping
Physical Channel: data rate r, time slot i

frequency

Logical Channel Mapping:

Different channels may share a same physical channel

frequency
Frame 8 Frame 9 Frame 10 Frame 11 Frame 12
Logical channel A: data rate r/3, time slot i, frame 3k
Logical channel B: data rate 2r/3, time slot i, frame 3k+1, 3k+2

Giuseppe Bianchi

GSM logical channels

Traffic channel (TCH) TCH/F TCH full rate MSÅÆBSS

TCH/H TCH half Rate MSÅÆBSS
Broadcast channel BCCH Broadcast control BSSÆMS
(same information to all MS in a cell) FCCH Frequency Correction BSSÆMS
SCH Synchronization BSSÆMS
Common Control channel (CCCH) RACH Random Access MSÆBSS
(point to multipoint channels) AGCH Access Grant BSSÆMS
(used for access management) PCH Paging BSSÆMS
Dedicated Control channel (DCCH) SDCCH Stand-alone Dedicated control MSÅÆBSS
(point-to-point signalling channels) SACCH Slow associated control MSÅÆBSS
(dedicated to a specific MS) FACCH Fast associated control MSÅÆBSS

Additional logical channels available for special purposes

(SMS, group calls, …)
Giuseppe Bianchi

13
An example procedure involving signalling

Setup for an incoming call (call

arriving from fixed network part -
MS responds to a call)

Steps:
- paging for MS
- MS responds on RACH
- MS granted an SDCCH
- authentication & ciphering on SDCCH
- MS granted a TS (TCH/FACCH)
- connection completed on FACCH
- Data transmitted on TCH

Giuseppe Bianchi

PART 4
GSM – Radio Interface

Lecture 4.3
Traffic channels and
associated signalling channels

Giuseppe Bianchi

14
 Traffic channels (TCH/F)
Periodic pattern of 26 frames (120 ms = 15/26 ms/TS * 8 TS/frame* 26 frame)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

24 TCH frames over 26

Same TS in every frame 01234567012345670123456701234567

Theoretical rate: 1/8 channel rate: r=33.85 kbps

2 signalling frames: r Æ 31.25 kbps TB DATA S Training S Data TB GP

3 57 1 Seq. (26) 1 57 3 8.25
Burst overhead (114 bits over 156.25):
148 bit burst
rÆ 22.8 kbps 156.25 bit (0.577 ms)
Giuseppe Bianchi

Speech coding

Analog A/D conversion Digital voice

voice 8000 samples/s 104 kbps
13 bit/sample

160 voice
samples Speech CODER 260 bits block
(20 ms) (8:1 compression) 13 kbps
(2080 bits)

Giuseppe Bianchi

15
 Discontinuous transmission
Î Speech coder implements Voice Activity Detection (VAD)
Ö Voice activity: idle for about 40% of the time
Ö To avoid clipping: hangover period (80ms)

talking listening talking

time
Î When IDLE, do not transmit
Ö Save battery consumption
Ö Reduces interference
Î Receiver side: silence is disturbing!
Ö Missing received frames replaced with “comfort noise”
Ö Comfort noise spectral density evaluated by TX decoder
Ö And periodically (480ms) transmitted in special frame (SID= Silence Descriptor)

Giuseppe Bianchi

Channel Coding
Coding: needed to move from 10-1 to 10-3 radio channel native BER
down to acceptable range (10-5 to 10-6) BER
260 bits
260 bits block divided into
182 bits 78 bits -Class I: important bits (182)
-Class Ia: Most important 50
Parity bits
Tail bits(0000) -Class Ib: Less important 132
-Class II: low importance bits (78)
50 bits 3 132 bits 4

Convolutional coding, r=1/2

378 bits 78 bits

456 bits
First step: block coding for error detection in class Ia (errorÆdiscard frame)
Second step: convolutional coding for error correction
Giuseppe Bianchi

16
 Block interleaving
8
0 1 2 3 4 5 6 7

8 9 10 11 12 13 14 15

16 17 18 19 20 21 22 23

.. .. .. .. .. .. .. ..
57
.. .. .. .. .. .. .. ..
.. .. .. .. .. .. .. ..
.. .. .. .. .. .. .. ..
448 449 450 451 452 453 454 455

B1 B2 B3 B4 B5 B6 B7 B8

8 blocks, each with 57 bits

Giuseppe Bianchi

Diagonal Interleaving
Block n-1 Block n Block n+1
B1 B2 B3 B4 B5 B6 B7 B8 B1 B2 B3 B4 B5 B6 B7 B8 B1 B2 B3 B4 B5 B6 B7 B8

B5n −1 / B1n … … B8n −1 / B4n B5n / B1n +1 … … B8n / B4n +1

PRICE TO PAY: delay!! (block spreaded over 8 bursts Æ 37 ms)

Inter-burst Interleaving
= Bxn −1 = Bxn− 4
Training
TB S S TB GP
sequence
3 1 26 1 3 8.25

Giuseppe Bianchi

17
 Bad Frame Indicator

ÎWhen errors cannot be corrected:

ÖFrame must be discarded
ÎIn substitution, transmit speech
frame
Öpredictively calculated
ÖNo more than 16 consecutive BFI
ÆIf this happens, receiver muted

Giuseppe Bianchi

Slow Associated Control Channel

Î Always associated to instaurated call on TCH (TCH + SACCH = TACH)
ÎOn the same Time Slot
Î Periodic (order of ½ second) time-scale information for radio link control
TCH/F(0…7) TCH/F(0…7)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

SACCH(0…7) IDLE frame

SACCH-0 SACCH-1 SACCH-2 SACCH-3 SACCH-4 SACCH-5 SACCH-6 SACCH-7

1 SACCH burst (per TCH) every 26 frames (120 ms)

Giuseppe Bianchi

18
 SACCH block
Î184 signalling bits
ÆBlock coding adds 40 bits (=224)
Æ4 tail (zero) bits (=228)
Æ1/2 Convolutional encoding (=456 bits)
ÎInterleaving: B1 B2 B3 B4 B5 B6 B7 B8
Ö 8 blocks of 57 bits;
Ö spreaded into four consecutive bursts
Ö4 bursts = 1 and only 1 SACCH block!
ÎSACCH rate: Odd/even
Ö184 bits/480 ms = 383.3 bit/s interleaving

104 frames = 480 ms

Giuseppe Bianchi

SACCH contents
Î 184 bits = 23 bytes
0 1 2 3 4 5 6 7
Î Power level
free Power level
Î Timing advance
free Timing advance
Î Measurement reports for
link quality
Î Measurement reports for (21 bytes – datalink layer)
handover management Includes measurement reports

Î When available space:

SMS
Ö When call in progress!

Giuseppe Bianchi

19
 Power control
Î MS has ability to reduce/increase
Maximum power power
(defined by class) Ö Up to its power class maximum
Î Maximum one 2 dB step every 60
ms
Î uplink power measures taken by
BTS
2 dB steps;Î notified back to MS via SACCH
Ö Power level values: 0-15
Æ0 = 43 dBm (20 W)
Æ15 = 13 dBm (20 mW)
Î algorithm: manifacturer specific
Ö runs on BSC
Minimum power
(13 dBm for GSM) Î power control applied also on
(0 dBm for DCS 1800) downlink
Giuseppe Bianchi

Measurement values
ÎRXLEV ÎRXQUAL
Ö Power level Ö Bit Error Rate (raw)
Ö Present channel + neighbohr cell)
Bit error From To
RX signal From To Ratio (%) (%)
level (dBm) (dBm) RXQUAL_0 - 0.2
RXLEV_0 - -110 RXQUAL_1 0.2 0.4
RXLEV_1 -110 -109 RXQUAL_2 0.4 0.8
RXLEV_2 -109 -108
RXQUAL_3 0.8 1.6
RXLEV_3 -108 -107
… … … RXQUAL_4 1.6 3.2

… … … RXQUAL_5 3.2 6.4

RXLEV_62 -49 -48 RXQUAL_6 6.4 12.8
RXLEV_63 -48 - RXQUAL_7 12.8 -

Averaged over 1 SACCH block (480ms = 104 frames)

Giuseppe Bianchi

20
 Fast Associated Control Channel
Î FACCH: urgent signalling
Ö Used when several signalling information needs to be transmitted
ÆCall setup
ÆHandover
Î FACCH block = 184
Ö 456 after coding
Î Interleaved as voice block
Ö Spreaded on 8 bursts
Î Replaces a voice block (20 ms) on the TCH
Ö Via stealing bits
Ö Voice block(s) discarded
Î Maximum FACCH bit rate
Ö 184*6/120 [bits/ms] = 9.2 kbps (vs 383 bps of SACCH!)

Giuseppe Bianchi

FACCH insertion in TCH

Via Stealing bits

- upper bit = odd bits stolen
- lower bit = even bits stolen
- both bits = all burst stolen

Figure: shows example of 2 FACCH blocks

stealing a TCH (note begin and end behavior
due to interleaving)

time
Giuseppe Bianchi

21
 Half-rate traffic channels (TCH/H)
ÎSupport for 5.6 kbps voice codecs
ÎSpecification in 1995
Î228 bits block
Ö112 bits of compressed 20ms voice
Æ95 bits class I
» + 3 parity + 6 tail + convolutional coding 104/211
Æ17 bits class 2
ÎInterleaving:
ÖSame as voice (block + diagonal + odd/even)
ÖBut on 4 bursts
ÎFraming:
TCH/H 0…7 TCH/H 8…15
[subchannel 0] [subchannel 1] SACCH 0…7 SACCH 8…15

1 2 3 4 5 6 7 8 9 10 11 12
Giuseppe Bianchi

Other traffic channels

ÎTCH data
ÖBasic speed: 9.6 kbps (data,fax)
ÖOther speeds: <2.4; 4.8; 9.6; 14.4;
ÎDifferent coding details
ÖSee text(s)
ÎMajor difference with voice:
ÖInterleaving with depth = 19 (!!!)
Öcomplete fading of a burst is recoverable (unlike
voice)

Giuseppe Bianchi

22
 PART 4
GSM – Radio Interface

Lecture 4.4
Broadcast Carrier and
Channels

Giuseppe Bianchi

Broadcast Channel (usual)

organization
Î 51 frame structure vs 26
Ö 235.38 ms period (vs 120 ms)
Î Sub-blocks with 10 frames
Ö Starting with Frequency Correction Channel (FCCH)
Ö Immediately followed by Synchronization Channel (SCH)
Î Other frames (numbered from #0 to #50):
Ö #50 idle
Ö #2,3,4,5 BCCH
Ö Remaining: Paging (PCH) / Access Grant (AGCH) [=PAGCH]

51 frame structure - downlink

FSBBBBPPPPFSPPPPPPPPFSPPPPPPPPFSPPPPPPPPFSPPPPPPPP

10 frame sub-block

Giuseppe Bianchi

23
 BCCH carrier placement
ÎOn Downlink
ÖCorresponding uplink dedicated to Random Access Channel
51 frame structure - uplink
RRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRR

ÎOn one frequency per cell (beacon)

ÎMUST BE on Time Slot #0
ÎOther Time slots may be used by TCH
Provided that:
• All empty slots are filled with DUMMY bursts
• Downlink power control must be disabled

Giuseppe Bianchi

MS powering up

First operation when MS turned ON: spectrum analysis

(either on list of up to 32 Radio Frequency Channel Numbers of current network)
(or on whole 124 carriers spectrum)

Giuseppe Bianchi

24
 tuning
ÎMS listens on strongest beacon for a pure
sine wave (FCCH)
ÖCoarse bit synchronization
ÖFine tuning of oscillator
ÎImmediately follows SCH burst
ÖFine tuning of synchronization (64 bits training sequence)
ÖRead burst content for synchronization data
Æ25 bits (+ 10 parity + 4 tail + ½ convolutional coding = 78
bits)
Æ6 bits: BSIC
Æ19 bits: Frame Number (reduced)
ÎFinally, MS can read BCCH

Giuseppe Bianchi

Multi-framing structure
8 TS, 4.615 ms
0 1 2 3 4 5 6 7

1 multiframe = 26 TDMA frames (120 ms) 1 multiframe = 51 TDMA frames (235.38 ms)

TTTTTTTTTTTTS TTTTTTTTTTTT FSBBBBPPPPFSPPPPPPPPFSPPPPPPPPFSPPPPPPPPFSPPPPPPPP

Multiframe 0 Multiframe 1 Multiframe 2 ……… Multiframe 49 Multiframe 50

Multiframe 0 Multiframe 1 … Multiframe 25

1 superframe = 51 x 26-multiframe or 26 x 51-multiframe (1326 TDMA frames, 6.12 s)

superframe 0 superframe 1 superframe 2 superframe 2046 superframe 2047

1 hyperframe = 2048 superframe (2715648 TDMA frames, 3h28m53s.76)

Giuseppe Bianchi

25
 Why frame number?

FN = Superframe # Multiframe # frame #

ÎFrame #
ÖDistinguishes logical channels in the same physical channel
ÎMultiframe #
ÖDetermines how BCCH is constructed
ÆI.e. which specific information transmitted on BCCH during a
given multiframe
ÎSuperframe #
ÖUsed as input parameter by encyphering algorithm

Giuseppe Bianchi

BCCH contents
Î 184 bits
Ö Coded in 456 bits and interleafed in 4 bursts
Ö same coding and interleaving as SACCH
Ö BCCH capacity
Æ184 bits / (51*8*15/26 ms) ~ 782 bps
Î Information provided
Ö Details of the control channel configuration
Ö Parameters to be used in the cell
ÆRandom access backoff values
ÆMaximum power an MS may access (MS_TXPWR_MAX_CCCH)
ÆMinimum received power at MS (RXLEV_ACCESS_MIN)
ÆIs cell allowed? (CELL_BAR_ACCESS)
ÆEtc.
Ö List of carriers used in the cell
ÆNeeded if frequency hopping is applied
Ö List of BCCH carriers and BSIC of neighboring cells

Giuseppe Bianchi

26
 control channel alternative
organization
51 frame structure – small capacity cell

DOWN FSBBBBPPPPFSPPPPPPPPFSDDDDDDDDFSDDDDDDDDFSDDDDDDDD

UP DDDDRRDDDDDDDDRRRRRRRRRRRRRRRRRRRRRRRDDDDDDDDRRDDDD

Integrates SDCCH in same channel as other control information

Leaves additional TS all available for TCH

51 frame structure – large capacity cell

DOWN BBBBPPPP PPPPPPPP PPPPPPPP PPPPPPPP PPPPPPPP

UP RRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRR

Used in TS 2 (and, eventually, 4 and 6) of beacon carrier

Provides additional paging and RACH channels
Giuseppe Bianchi

Why 26 and 51?

Last frame (idle) in TCH multiframe (Frame #25) used as “search frame”!

TTTTTTTTTTTT S TTTTTTTTTTTT

- An active call transmits/receive in 25 frames, except the last one.

- in this last frame, it can monitor the BCCH of this (and neighbor) cell
- this particular numbering allows to scan all BCCH slots during a superframe

- important slots while call is active: frequency correction FCCH and sync SCH!
- needed for handover
-Worst case: at most every 11 TCH multiframes (1.32 s), there will be
a frequency correction burst of a neighboring cell

Giuseppe Bianchi

27
 PART 4
GSM – Radio Interface

Lecture 4.5
Paging, Random Access,
dedicated signaling

Giuseppe Bianchi

Why paging
ÖChannel assignment:
Æonly upon explicit request from MS
ÖPaging
Æneeded to “wake-up” MS from IDLE state when incoming call
arrives to MS
ÖMS accesses on RACH to ask for a channel
ÆGenerally SDCCH (but immediate TCH assignment is possible)
1) paging

MS 2) Random access BSS/MSC

3) Channel assignment

Paging channel: PCH CCCH

PAGCH
Access Grant Channel: AGCH Common Control
Random Access Channel: RACH CHannel
Giuseppe Bianchi

28
 Paging
ÎPaging message generated by MSC
ÖWhich receives incoming call
ÎTransferred to subset of BSC
ÖPaging limited to user’s location area
ÖPaging message contains:
ÆList of cells where paging should be performed
ÆIdentity of paged user (IMSI or TMSI)
ÎPaging message coded in 4 consecutive
bursts over the air interface
ÖSame coding/interleaving structure of SACCH (184Æ456 bits)
ÎPaging for more MSs may be joined in one
unique paging message

Giuseppe Bianchi

Discontinuous Reception (DRX)

Î MS in idle mode should listen to paging channel
Î To save battery, PCH divided into sub-channels
Ö Subdivision based on last 3 digits of IMSI
Ö User listens only to relevant sub-channel
ÆSwitches off otherwise
Î PCH sub-channels pattern communicated on
BCCH
Ö Up to 9 sub-channels
51 frame BCCH structure - downlink
FSBBBBPPPPFSPPPPPPPPFSPPPPPPPPFSPPPPPPPPFSPPPPPPPP

Giuseppe Bianchi

29
 Access procedure
Î Always activated by MS
Ö In response to paging (incoming call)
Ö When location update needs to be performed
Ö When new call is generated by MS
Î Based on access burst sent on RACH
Ö Access burst coding: 8 payload bits (channel_req) Æ 36 coded bits
Æ+6 parity; + 4 tail; + ½ convolutional coding

Access burst
TB Training Data TB GP
8 Seq. (41) 36 3 68.25

100: response to radio call

101: emergency call Channel_req message
110: new establishment of call Establishment Random discriminator
111: supplementary service (SMS, etc) Cause (3 bits) (5 bits)
…
000: other case
Random discriminator: (0…31) value randomly generated by MS
Giuseppe Bianchi

RACH operation
MS-C

A,B C,B C A,B A B

MS-A MS-B
Multiple Access Technique for simultaneous access
Collision resolution based on
- random retrial period
- “permission” probability Same thing..!
(SLOTTED ALOHA protocol)
Giuseppe Bianchi

30
 RACH performance
ÎN stations
ÎEach transmits with probability p
ÖRetries, in average, every 1/p slots
Îrelevant probabilities:
idle : (1 − p )N
Np (1 − p )
N −1
success :
N
⎛N⎞
∑ ⎜⎜ k ⎟⎟ p (1 − p )
k N −k
collision :
k =2 ⎝ ⎠
Maximum efficiency: when p=1/N

Giuseppe Bianchi

RACH performance control

ÎBCCH broadcasts
ÖBackoff time
Æuniform distribution; max value: 3 to 50
ÖMaximum number of retransmission attempts
ÆNever greater than 7
ÎMay also specify:
Ötime interval for a retry after failure (default=5s)
ÖRACH group access control
ÆMSs divided into 10 groups, depending on SIM-related
information
ÆBTS may block selected groups
ÆAllows to reduce RACH load down to as low as 10%
» Emergency calls bypass this rule…

Giuseppe Bianchi

31
 Access signaling - 1
MS BTS BSC MSC
Channel_request
rnd number
Channel_required
rnd&frame number,
Delay (TA estimate)

Channel_activation

Ch_activation_ack

Immediate_assignment
rnd&frame number, channel description,
Initial TA, initial max power

Giuseppe Bianchi

Immediate Assignment message

Îon paging channel
ÖI.e. AGCH is a dynamically mapped channel
– name PAGCH is perhaps better…

ÎSent “as soon as possible”

ÖMS continues accessing the RACH
ÖMessage scheduling is an implementation dependent issue
ÆI.e. which message to send in case of many messages, and on
which paging slot (4 bursts)
ÖMS must disable DRX
ÆTo monitor PAGCH for Immediate Assignment message
detection
ÎImmediate assignment reject possible

Giuseppe Bianchi

32
 If same random discriminator?
Î two MSs may have same random discriminator
Ö Likely in heavy load, with only 5 bits
Î And transmit in the same frame
Î Only one wins (the other is faded)
Î contention resolved via explicit MS identification
Ö On SDCCH MS1 MS2 BTS
ID1

ID2

ID1
leave

continues
Giuseppe Bianchi

Access signaling - 2
MS BTS BSC MSC
… … … …
Immediate_assignment
Initial_message

MS ID (IMSI or TMSI), MS capabilities

(=classmark), establishment cause

Initial_message_ack (UA) Establishment_indication

Copy of Initial message (including MS ID)

Further signaling: MSC to MS

Giuseppe Bianchi

33
Stand-alone Dedicated Control
Channel - SDCCH
Î Dedicated bidirectional channel to MS
Ö but used to exchange signalling
Î Has an associated SACCH
Î Coding: as SACCH
Ö 184Æ456 bits on 4 consecutive bursts
SDCCH/4 for small
Î Typical framing (SDCCH/8)
cells – SDCCH shares
Ö 8 SDCCH (+8 SACCH) on 1 channel (carrier,TS) BCCH+PAGCH channel
Ö 1 SDCCH message per 51-multiframe - see before -
Æ184 bits / (51*8*15/26 ms) = 598/765 kbps ~ 782 bit/s
Ö 1 SACCH every 2 multiframe
DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDAAAAAAAAAAAAAAAA
000011112222333344445555666677770000111122223333

DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDAAAAAAAAAAAAAAAA
000011112222333344445555666677774444555566667777

Giuseppe Bianchi

34
 PART 5
GSM – Switching & Mobility

Lecture 5.1
Protocol architecture overview

Giuseppe Bianchi

The GSM network layer

Î Divided in three sub-layers
Ö Radio Resource Management (RR)
ÆProvides a communication link
between MS and MSC;
Ö Mobility Management (MM)
CM
ÆManages DB for MS location
Ö Communication Management (CM)
MM
ÆControls user connection
RR
Î Underlying base:
Transmission
Ö Transmission level

Giuseppe Bianchi

1
 RR
Î Manages administration of frequencies and channels
Ö Mostly deals with air interface
ÆSeveral RR functions considered in previous part
Î Guarantees stable link upon handover
ÆSurprise! handover is part of RR, not MM!
Î Function summary:
Ö Monitoring BCCH, PCH
Ö RACH administration
Ö Request/assignment of channels
Ö MS power control & synchronization
Ö Handover
Î Where is RR:
Ö MS, BTS, BSC, MSC

Giuseppe Bianchi

MM
Î Manages user location and tasks resulting from
mobility
Î Function summary:
Ö TMSI assignment
Ö MS localization
Ö Location updating
Ö MS authentication
Ö MS identification, attach/detach
Î Where is MM:
Ö MS, MSC

Giuseppe Bianchi

2
 CM
Î Controls calls, supplementary services, and SMS
Î Function summary:
Ö Call establishment (from MS, to MS)
Ö Emergency call management
Ö Call termination
Ö DTMF signaling (Dual Tone MultiFrequency)
Ö In-call modification
Î Where is CM:
Ö MS, MSC, GMSC

Giuseppe Bianchi

Protocol placement

CM

GMSC
MM

HLR
RR

Trans.

MSC
MS BTS BSC
(VLR)

Giuseppe Bianchi

3
 Protocol outline
Relay Anchor
MS BTS BSC MSC MSC HLR

CM RIL3-CC
MAP/D
MM RIL3-MM

RR RIL3-RR
RSM BSSMAP MAP/E

TCAP

SCCP SCCP SCCP

LAPDm LAPD MTP MTP MTP

RIL3: Radio Interface Layer 3 TCAP: Transaction Capabilities Application Part

RSM: Radio Subsystem Management
BSSMAP: BSS Management Application Part
MAP: Mobile Application Part
Giuseppe Bianchi
SCCP: Signaling Connection Control Part
MTP: Message Transfer Part
LAPD: Link access Protocol on D channel
LAPDm: Link access Protocol on Dm channel

PART 5
GSM – Switching & Mobility

Lecture 5.2
handover (physical mobility)

Giuseppe Bianchi

4
 Neighbor cells
BTS
2

BTS BTS
1 n

Î A station must:
Ö monitor beacon power level of neighbor cells
Ö Keep detailed track of best 6 neighbor cells
Ö DECODE their BCCH (i.e. read FCCH, SCH) to get parameters
ÆAt least once every 5 minutes
ÆBSIC (from SCH) refreshed every at most 30s

Giuseppe Bianchi

Camping cell selection

path loss criterion C1
Select cell with greatest c1(n)>0:
C1(n) = RXLEV(n) −
− RXLEV_ACCESS_MIN −
− max[0, (MS_TXPWR_MAX_CCH − P )]

Î RXLEV(n): received power from BTS(n)

Î RXLEV_ACCESS_MIN: minimum received power level required for registration in the cell
Î (parameter transmitted on BCCH; typically –98 to –106 dB)
Î MS_TXPWR_MAX_CCH: maximum allowed transmitted power on RACH
Î (parameter transmitted on BCCH; typically 31-39 dBm)
Î P: maximum MS power (from MT class)

When cell parameters are the same, simply select cell with higher RXLEV!

Giuseppe Bianchi

5
 Cell reselection criterion (C2)
ÎReselect cell with greatest C2>0:

C2(n) = C1(n) + CELL_RESELECT_OFFSET −

− TEMPORARY_OFFSET × H(PENALTY_TIME − T)
⎧0 x < 0
where H(x) = ⎨
⎩1 x ≥ 0
Î T: amount of consecutive time since considered cell became with C1>0
Î PENALTY_TIME, CELL_RESELECT_OFFSET, TEMPORARY_OFFSET: BCCH parameters
Î If all parameters = 0, reselect cell with better path loss performance (no time hysteresis included)

Giuseppe Bianchi

Consequences of cell reselection

BTS BTS

Î None, when MS idle!

ÆNo need to inform BTS at all!
Î Exception:
Ö When cell reselection implies a Location Area Update
ÆNeed to inform the network!
Î Additional restriction:
Ö C2>CELL_RESELECT_HYSTERESIS
Giuseppe Bianchi

6
 handover
Î Procedure in which an MS releases a connection with
a BTS, and establishes a connection with a new BTS,
while ensuring that the ongoing call is maintained
Ö The MS remains in dedicated state (unlike cell reselection, where MS is in idle
state)
Î Handoff: synonymous of handover
Î Needs two mechanisms
Ö Handover preparation: detection of cell-border crossing
ÆBased on radio link quality measurements
Ö Handover execution: setup of a new channel in a cell, and tear-down of a
previous channel
Î Improved handover mechanisms:
Ö Seamless handover: when active call performance is not impaired
ÆNot possible in GSM: for about 100-200ms, communication is interrupted
Ö Soft Handover: when two channels are simultaneously set-up (old and new)
ÆNot possible in GSM; possible in UMTS

Giuseppe Bianchi

Hard, Seamless, Soft handover

before during after
MSC MSC MSC
Hard
handover BSS 1 BSS 2 BSS 1 BSS 2 BSS 1 BSS 2
(GSM) f1 f1 f2
MS MS MS

MSC MSC MSC

Seamless
BSS 1 BSS 2 BSS 1 BSS 2 BSS 1 BSS 2
(DECT)
f1 f1 f2 f2
MS MS MS

MSC MSC MSC

Soft
handover BSS 1 BSS 2 BSS 1 BSS 2 BSS 1 BSS 2
(UMTS) f1 f1 f1 f1 f1
MS MS MS

Giuseppe Bianchi

7
 Handover classification
Classification by motivation Classification by typology

Î Rescue handover Î Internal handover

(mandatory handover)
Ö Driven by radio channel quality
degradation
Î Confinement handover
(network-directed handover)
Ö Target: minimize radio interference
Ö Assign new channel when old
channel results critical for total
interference
Î Traffic handover
(network-directed handover)
Ö Driven by traffic congestion
conditions
Ö Also called load-balancing
Giuseppe Bianchi
Ö Intra-BTS
ÆNew radio channel in the same
cell
ÆNot termed as “handover” but
as“subsequent assignment”
Ö Inter-BTS (Intra-BSC)
ÆUnder control of same BSC
Î External handover
Ö Inter-BSC (Intra-MSC)
ÆChange reference BSC; may
imply a location area update
Ö Inter-MSC
ÆMost complex: need to change
MSC

Types of handover
Anchor MSC: Relay MSC:
the MSC that first the MSC that currently
managed the current call manages the call
Switching Switching
point for point for
all inter-MSC A-MSC R-MSC
inter-BSC
handover handover
A
Switching
point for
internal BSC BSC BSC
handover
A-bis

BTS BTS BTS BTS

radio
interface

Giuseppe Bianchi

8
 Handover taxonomy
Î BCHO: Base station Controlled Handover
Ö Handover detection: BS
Ö Handover Execution: BS
Î MCHO: Mobile Controlled Handover
Ö Handover detection: MS
Ö Handover Execution: MS
Î MAHO: Mobile Assisted Handover
Ö Handover detection: MS
Ö Handover Execution: BS

Î GSM: somehow a BCHO with a flavor of MAHO

Ö Handover decision always taken by BSC
Ö Based on measures taken at both BTS and MS
Ö New channel selection decision taken at BSC or R-MSC or A-MSC
(depending on handover type) based on traffic consideration

Giuseppe Bianchi

Handover preparation RX signal

level
From
(dBm)
To
(dBm)
Î Measurements performed at BTS RXLEV_0 - -110
Ö Up-link signal level received from MS lower than threshold RXLEV_1 -110 -109

ÆRXLEV_UL < L_RXLEV_UL_H RXLEV_2 -109 -108

Ö Up-link signal quality (BER) received from MS RXLEV_3 -108 -107

… … …
ÆRXQUAL_UL < L_RXQUAL_UL_H
… … …
Ö Distance between MS and BTS RXLEV_62 -49 -48
Æadaptive timing advance parameter > MAX_MS_RANGE RXLEV_63 -48 -
Ö Interference level in unallocated time slots.
Î Measurements performed at MS.
Bit error From To
Ratio (%) (%)
Ö Down-link signal level received from serving cell RXQUAL_0 - 0.2
ÆRXLEV_DL < L_RXLEV_DL_H RXQUAL_1 0.2 0.4

Ö Down-link signal quality (BER) received from serving cell RXQUAL_2 0.4 0.8

ÆRXQUAL_DL < L_RXQUAL_DL_H RXQUAL_3 0.8 1.6

Ö Down-link signal level received from n-th neighbor cell RXQUAL_4 1.6 3.2

ÆRXLEV_NCELL(n) > RXLEV_MIN(n) RXQUAL_5 3.2 6.4

RXQUAL_6 6.4 12.8

RXQUAL_7 12.8 -
Giuseppe Bianchi

9
 A note on MS distance

ÎDistance can be measured based on TA

ÎTA = advance bits
ÖIdeally, TA should be set as

TA[bits ]⋅ tbit =
2d TA
⇒ d= ⋅ c ⋅ tbit
c 2
ÖHence, the TA resolution, in mt, is:
1
300000[mt / ms] ⋅ [ms ]
c ⋅ tbit
d (TA) = TA = TA 270.833 ≈ TA ⋅ 554mt
2 2
ÖINSUFFICIENT for microcells!
ÖSufficient only to understand we are going out of the cell
Giuseppe Bianchi

Handover preparation –
additional metrics
Î Transmission power
Ö Maximum MS transmission power
RXLEV RXLEV
Ö Maximum serving BTS transmission power (cell A) Handover (cell B)
Ö Maximum neighboring BTSs transmission
power
Î congestion status
Ö of serving BTS
Ö of neighboring BTSs
Æ provided they can support the MS. RXLEV RXLEV
Î Handover Margin (cell A)
Handover
(cell B)
Ö To avoid ping-pong handover effect
Ö 5-10 dB in normal operation; up to 30dB in
urban operation (to fight shadowing)

HANDOVER ALGORITHM: operator-dependent!

GSM standard SUGGESTS a simple reference
algorithm, but implementation left to operator hysteresis

Giuseppe Bianchi

10
 handover procedure skeleton
1) Handover request goes up to switching point

MSC
2) Switching point prepares new path on fixed net

3) Switching point sends HO command to MS

2
1
5
4) MS accesses new channel
BSC
BSC
5) Old channel/path torn down
3
4 BTS
BTS

Giuseppe Bianchi

Signaling for intra-MSC handover

(simplified)
MS BTS-A BSC-A MSC BSC-B BTS-B MS

Measurement info handover required

(destination cell)
handover request
Channel allocation
Channel activation
handover req. ack ACK
(contains handover
handover command command message
handover command
prepared by BSC-B
with info on BCCH,
channel assigned, etc)
handover access
handover detection (an access burst
handover detection on new TCH!!!)
physical info
(new TA, power)
handover complete
handover complete
clear command

clear complete Measurement

info

Giuseppe Bianchi

11
 Inter-MSC handover
Î More complex, as an ISDN circuit must be set
between MSCs
Ö We’ll not enter into details (just the basic ideas)
Î Two cases

First MSC change Second MSC change

(basic handover) (subsequent handover)

MSC-A MSC-R1 MSC-A X MSC-R1 MSC-R2

X Note the role of the X

Anchor MSC!

Giuseppe Bianchi

PART 5
GSM – Switching & Mobility

Lecture 5.3
location registration/update
Authentication & Ciphering

Giuseppe Bianchi

12
 Location Area
vs
MSC service area

LA-1 LA-2 LA-3

MSC VLR
LA-4 … LA-n

Giuseppe Bianchi

Registration vs update
Î Very similar procedures, with goals:
Ö Determine where the user is
Ö Authenticate user
Î Differences:
Ö Location Registration
ÆUser first access to PLMN
» Needs to send IMSI and receive TMSI
Ö Location Update
ÆSubsequent accesses to PLMN (either in old or new MSC/VLS)
» Also after MS shut-down!
» TMSI-based identification
Î Registered user:
Ö The PLMN knows the LA where the user is (or is supposed to be)

Giuseppe Bianchi

13
 Procedure start-up

ÎMS switches on
ÎDetects BCCH carrier
ÖTune and synchronize
ÎListens to BCCH
ÎObtains Location Area Identifier
ÖLAI: [CC,MNC,LAC]
ÆCountry Code (CC): 3 digits
ÆMobile Network Code: 2 digits
ÆLocation Area Code: max 5 digits

Giuseppe Bianchi

LR/LU (very) basic idea

3
MSC VLR HLR

BSC
BTS
2 1) Obtain LAI from BCCH
BTS
2) Register MS ID into local VLR
1 BTS
3) Update pointer at HLR
MS

Giuseppe Bianchi

14
 Location Registration
MS BSS/MSC VLR HLR AUC
Loc. Upd. Request
Update Loc. Area
IMSI, LAI Auth. Param. Req. Auth. Info. Req.
IMSI, LAI
IMSI IMSI
Auth. Info Auth. Info
(Auth. Parameters) (Auth. Parameters)

authentication
Update Location
Start Ciphering IMSI, MSRN
Kc Insert Subscrib. Data
Activate Forward new TMSI IMSI, additional data
ciphering TMSI Insert Subscrib. Data
ACK
Locat. Upd. Accept
Locat. Upd. Accept
IMSI

TMSI Realloc Cmd

Locat. Upd. Accept
TMSI Realloc ACK
TMSI ACK
Giuseppe Bianchi

Authentication
(managed by VLR)

IMSI, RAND HLR /

Authentication Request VLR
AUC
MS SRES, Kc
Challenge: 128 bit RAND
Ki RAND

A3 SRES

SRES
Authentication Response
Equal?
Signed RESult: 32 bit SRES
Ki RAND
A8

Kc
Giuseppe Bianchi

15
 Authentication (details)
Î Side effect of authentication:
Ö Generate encryption key Kc via A8 algorithm
Î Secret A3, A8 algorithms (one-way hash functions)
Ö Stored into the SIM
ÆAlong with secret key Ki
Ö Note that roaming operator DOES NOT need to know them!
ÆSince A3,A8 run ONLY in the AUC at the home HLR
ÆKi is NEVER transmitted away from AUC or MS!
Î Generally implemented together
Ö [SRES,Kc] = A38[Ki,RAND]
Î To reduce signaling, real implementation slightly
different:
Ö VLR sends IMSI
Ö Receives back several tuples of (RAND, SRES, Kc) to be used for the
considered MS also in subsequent accesses
Giuseppe Bianchi

ciphering
Î A5 algorithm is known (to allow roaming)
Î Generates two ciphering sequences
Ö one for uplink, one for downlink
Ö Sequence periodic with period 26x51x2048=2,715,648
Æ 221=2,097,152 < 2,715,648 < 222=4,194,304
Î 114 bits per frame, depending on frame number
Î XOR-ed with burst data field

Frame number Kc Frame number Kc

FN, 22 bits 64 bits FN, 22 bits 64 bits

MS A5 A5 BTS
S1 S2 S1 S2

In-clear uplink XOR enciphered uplink XOR In-clear uplink

In-clear downlink XOR enciphered downlink XOR In-clear downlink

Giuseppe Bianchi

16
 Location Update in same VLR
(same as location registration, but with TMSI)
MS BSS/MSC VLR HLR AUC
Loc. Upd. Request
Update Loc. Area
TMSI, LAI
TMSI, LAI
authentication Update Location
IMSI, MSRN
Generate
New TMSI
Start Ciphering
Activate Kc Insert Subscrib. Data
ciphering Forward new TMSI IMSI, additional data
Ins. subs. data ACK
Locat. Upd. Accept
Locat. Upd. Accept
IMSI
TMSI Realloc Cmd
Locat. Upd. Accept
TMSI Realloc ACK TMSI ACK
Auth. Param. Req. Auth. Info. Req.
IMSI IMSI
Auth. Info Auth. Info
N x (Kc,RAND,SRES) N x (Kc,RAND,SRES)
Giuseppe Bianchi

Changing MSC/VLR

HLR
VLR MSC Public
Publicswitched
switched
telephone
telephonenetwork
network
PSTN
PSTN

VLR MSC

Base
Station

Base
Station

An MS always has a dedicated entry in the HLR

Plus one entry in JUST 1 VLR
(related to the MSC the user is connected to)
Giuseppe Bianchi

17
 TMSI
ÎTMSI = Temporary Mobile Subscriber Identity
Ö4 octets (32 bits) Operator may set a 6min
ÖRenewed periodically; at every LU / IMSI_attach up to 24hrs periodicity
for LU (value transmitted
ÆVia TMSI_Reallocation_Command/TMSI_Reallocation_Complete on BCCH)
ÆRATIONALE: renew TMSI when transmitted in clear!
IMSI_attach = a special LU
(TMSI reallocation occurs in ciphering mode)
in a same Location Area;
ÎMeaningful only in a given VLR IMSI_attach follows
ÖSpecifically, only for a given Location Area!! an IMSI_detach
(power-down of MS)
ÆSome author (Mouly-Pautet) uses the term
» TIC (Temporary Identity Code) = 4 bytes
» TMSI = TIC+LAI = unambiguous user identification
ÎWhile entering a new Location Area:
Ö user must identify itself with TMSI+LAI pair.

Giuseppe Bianchi

Location Update: different VLR

MS BSS/MSC VLR-new HLR VLR-old
Loc. Upd. Request
Update Loc. Area
TMSI(+ old LAI), LAI
TMSI(+ old LAI), LAI
determine
VLR-old
From old LAI Send parameters (TMSI, old LAI)

IMSI response (IMSI,RAND,SRES,Kc)

authentication
Update Location
Cancel Location
IMSI, MSRN
Generate IMSI
New TMSI Cancel Locat. ACK
Start Ciphering
Activate Kc Insert Subscrib. Data
ciphering Forward new TMSI IMSI, additional data
Ins. subs. data ACK
Locat. Upd. Accept
… …
IMSI

Giuseppe Bianchi

18
 Special cases
1. New VLR not capable of determining old
VLR from old LAI
2. Old VLR does not recognize TMSI
MS MSC
Identity Request
Ö Identification procedure
Æ IMSI transmitted in clear Identity Response
IMSI

PAGING:
- Normally based on TMSI
- But when no valid TMSI information available (e.g. after a DB restore
after crash), based on IMSI

Giuseppe Bianchi

PART 5
GSM – Switching & Mobility

Lecture 5.4
Call Management & routing

Giuseppe Bianchi

19
 Notation
ÎA call involves two “Parties”
ÎCalling Party (caller)
Öuser generating the call
ÎCalled Party (callee)
Öuser receiving the call
ÎMobile Originating Call (MOC)
ÖCall originated by an MS
ÎMobile Terminating Call (MTC)
ÖCall directed to an MS

Giuseppe Bianchi

Call establishment basics

Mobile Terminated Call Mobile Originated Call
Fixed Fixed
MS MSC party MS MSC party
setup
setup setup
setup Call proceeding
Call confirmed alerting
alerting
alerting
alerting Connect
Connect
connect
Connect Ack
connect

DATA DATA

In ISDN ISUP: - setup = IAM (Initial Address Message);

- Alerting = ACM (Address Complete Message);
- Connect = ANS (Answer)

Giuseppe Bianchi

20
 Call establishment steps
MS Mobile network MS Mobile network
Terminated Call Originated Call
Paging request
Channel request Channel request
Immediate Assignment Immediate Assignment
Paging Response Service Request
Authentication Request Authentication Request
Authentication Response Authentication Response
Ciphering mode command Ciphering mode command
Ciphering Mode Complete Ciphering Mode Complete
Setup Setup
Call Confirmed Call proceeding
Assignment Command Assignment Command
Assignment Complete Assignment Complete
Alerting Alerting
Connect Connect
Connect Acknowledge Connect Acknowledge

Giuseppe Bianchi

Radio Resource allocation

three standardized solutions

Î Non-Off Air Call Set-Up (Non-OACSU)

Ö Normally used (previous description)
Î Off Air Call Set-Up (OACSU)
Ö TCH assigned only when the called party actually responds!
ÆBest utilization of radio resource (avoids allocation if callee not available)
ÆCall drop if no TCH is available at this point
Î Very Early Assignment (VEA)
Ö Immediate assignment of TCH
ÆFastest signalling process
ÆWaste of resources

VEA RACH TCH (FACCH) TCH (DATA)

Non-OACSU RACH SDCCH TCH (FACCH) TCH (DATA)
OACSU RACH SDCCH TCH (DATA)

Connection established Callee responds

Giuseppe Bianchi

21
 DTMF signaling
Î Dual-Tone Multi-Frequency On air interface:
Ö Digital tones associated to terminal keys Ö Signal trasmitted on FACCH as
Æ‘0’…’9’…’#’… signalling data (code of pressed
Î Inband signalling key)
Ö transmitted in the traffic channels! Ö Otherwise coded compression
would distort DTMF tones
Ö Not in the signalling network
Ö Tone generated at MSC when
STOP DMTF message received
MS MSC
Start DTMF (w. key code)
On FACCH)
Key Start DTMF ACK
Pressed
Stop DTMF

Giuseppe Bianchi

Routing an MTC

ISDN
N
SD
SI
M
1:

PLMN
GMSC
2: 3: M
M
N

MSCC
SI SRN
SR

SD
M

N
4:

5: MSRN
VLRB MSCB HLR
6: TMSI MSCA
7:
pa
gi
ng

Giuseppe Bianchi

22
 Routing an MTC (alternative)
reduces signalling load during LU

During an LU within a same VLR,

MSRN is NOT signaled!
ISDN

N
D
MSRN retrieved on a per-call basis!

S
SI
(choice of solution depends on trade-offs)

M
1:
PLMN
GMSC

2: 5: M
M
N
MSCC

SI SRN
SR

SD
M

N
6:

7: MSRN
VLRB MSCB HLR
8: TMSI MSCA
9:
pa
gi

3: IMSI
ng

4: MSRN
Giuseppe Bianchi

Routing calls to Roaming MS

ISDN International
MSC
(UK) Switching
Center PLMN 2
(UK)

International
Switching
ISDN Center
(ita)
Transit
Exchange
GMSC 1 PLMN 1
(ITA)
Local MSISDN MSRN MSC
Exchange +39.335.1234567 +44.NDC.8877665

HLR
335.1234567

Giuseppe Bianchi

23
 “tromboning”
Call to MSISDN
+39.335.3043125

MSC
ISC
(UK) PLMN 2
(UK)

Call to MSISDN
+39.335.1234567

Is the PRICE (!)

ISC
GMSC 1 PLMN 1 to pay for
(ITA) (ITA) simple routing
and billing
MSISDN MSRN MSC
+39.335.1234567 +44.NDC.9876543

HLR

Giuseppe Bianchi

Tromboning technical solutions

ÎFirst alternative: national-wise

ÖAdd a new database - Roamer Location Cache (RLC)
ÆConsulted by ISCs (which MUST support GSM-MAP!)
ÎSecond alternative: PLMN specific
ÖRLC within the PLMN + associated switch
ÖCaller must dial special NDC number (the switch!)
ÆI.e. must know the MS is roaming in the PLMN…
ÖAdditional devices and protocol modifications required
» Extensions toVLR or to GMSC
» Details in “Lin-Chlamtac”

Giuseppe Bianchi

24
RLC at ISC - Location Registration
and call management

PLMN 2
ISC
RLC 3 bis (UK) 2 VLR MSC (UK)

PLMN 1
ISC (ITA)
(ITA) 4 HLR

Giuseppe Bianchi

Short Message Service

ÎSMS:
Ömessages up to 160 bytes
ÖMessage concatenation allowed
ÎTransmitted on air interface over:
ÖSACCH (when user in conversation)
ÖSDCCH (when user in idle state)
ÎTwo transmission modes in a cell:
ÖPoint-to-point
Öcell broadcast
ÎConnectionless service
Ömessage switching (store&forward)
ÖImplemented through the Short Message Service Center

Giuseppe Bianchi

25
 SMS routing management

MSC PLMN
Internet, PSDN

IWMSC

Short Message
Get routing info
Service Center for terminating MS

SMS-GMSC HLR

PLMN
MSC

Giuseppe Bianchi

Protocol hierarchy
MS
SM-SC
Short Message
Application Layer
(SM-AL)
Short Message Transfer Protocol (SM-TP)
Short Message
Transfer Layer
(SM-TL) MSC IW-MSC
Short Message Short Message Short Message
Relay Layer Relay Entity Short Message Relay Entity
(SM-RL) (SMR) Relay Protocol (SMR)
(SM-RP)
Connection Short Message Short Message Short Message
Management Control Entity Control Protocol Control Entity
Sublayer (CM-sub) (SMC) (SM-CP) (SMC)

Quite complex signalling involved (see specific texts)

Giuseppe Bianchi

26
 Number portability
ÎSubscriber may switch operator without
changing his number
ÎFirst implemented in fixed network
ÖRecently (may 2002) extended to mobile networks
ÎEssential for fair competition among
network operators
ÖUK survey: 42% of corporate subscribers were willing to
change mobile operator; but 96% were, if number could be
ported
ÎResistence from leading operators
ÖNumber portability helps newer operators to compete with
traditional ones

Giuseppe Bianchi

Notation

ÎDonor switch
ÖThe switch corresponding to a “ported” telephone
number
ÎRecipient switch
ÖThe switch to which the ported number is attached

Giuseppe Bianchi

27
 Technical solutions
a) call forwarding
Originating network Donor network

switch switch

switch
Recipient network
Originating switch sets-up trunk to donor switch
Donor switch sets-up trunk to recipient switch
Simplest solution, as call forwarding is a feature available in virtually all switches
But extremely inefficient routing and trunking resource consumption!
Giuseppe Bianchi

Technical solutions
b) query on release
Originating network Donor network
SS7 ISUP IAM
switch switch
SS7 ISUP REL

Number
Portability switch
DataBase
Recipient network

Donor switch “blocks” incoming call with a release message (REL)

REL carries a QoR cause value, stating that called party number is ported
Originating switch then queries Number Portability database

Giuseppe Bianchi

28
 Technical solutions
c) all-call query
Originating network Donor network

switch switch

Number
Portability switch
DataBase
Recipient network

Originating switch queries Number Portability database for every call!!

- best solution if majority of numbers are ported (no interaction with donor)
- but very high DB load, as EVERY number must be looked-up!

Giuseppe Bianchi

Mobile Number Portability

Î Same ideas as fixed number portability
Ö The donor switch is the GMSC of the donor network
Î Donor GMSC Call forwarding (if more efficient
fixed number portability not supported)
Ö While porting number, may also get MSRN!
Recipient network

MSC GMSC HLR Note: If path must cross GMSC:

Use Intermediate Routing Number
MSRN IRN MSRN
(or IRN)
Incoming call
Signaling relay
GMSC function HLR
Donor network
Clearly, still suffers of tromboning!
Giuseppe Bianchi

29
 Mobile Number Portability
(with all call query approach)
approach)

Recipient network
Query MSRN
MSC GMSC Return MSRN HLR

IRN

Incoming call
Query IRN Number
switch Return IRN Portability
DataBase

Giuseppe Bianchi

Mobile Number Portability

improved – (with all call query approach)
approach)

Recipient network

MSC GMSC HLR

MSRN
Signaling relay
function
Incoming call
Query MSRN Number
switch Return MSRN Portability
DataBase

Giuseppe Bianchi

30